feat(R2-Q4): adapter cutover — SetCycle/PlayAction rehosted on PerformMovement; Fix B / fast-path / stop-anim fallback / G17 gate DELETED

AnimationSequencer becomes a thin shim over the verbatim R2 stack:
SetCycle dispatches the style change (Branch 1) then the motion (Branch
2/3/4) through MotionTableManager.PerformMovement; PlayAction is the
same dispatch (actions rebuild via Branch 3; modifiers are Branch 4
physics-only combine_motion — the AP-73 turn-blend mechanism now runs
for real). CurrentStyle/CurrentMotion/CurrentSpeedMod are read-only
mirrors of MotionState (post-adjust_motion, signed mod). Lazy
initialize_state on first drive (retail lazy-create analog) keeps
undriven sequencers do-nothing.

DELETED legacy inventions (net -648 lines): the adapter fast-path
(replaced by Branch-2 fast re-speed: change_cycle_speed +
subtract/combine_motion), Fix B + IsLocomotionCycleLowByte (replaced by
remove_redundant_links on the pending queue — the retail mechanism the
2026-05-03 cdb trace pointed at), the stop-anim low-byte fallback
(replaced by get_link's reversed-key double-hop — retail's actual
backward-walk settle plays the windup link REVERSED), GetLink/BuildNode/
EnqueueMotionData + the G17 HasVelocity/HasOmega gate (add_motion sets
unconditionally), MultiplyCyclicFramerate + the G13 composite,
PlayAction's insert-before-tail machinery, SCFAST/SCFULL diagnostics.
KEPT at the adapter (register rows): the boundary adjust_motion remap +
locomotion velocity/omega synthesis (AP-75, retire R3-W6/R6), K-fix18
skip-link as post-dispatch RemoveAllLinkAnimations (AP-74, retire
R3-W4 when LeaveGround fires it).

GameWindow queue-drain wiring (the §4 G6 seam): each drained AnimDone
hook → manager.AnimationDone(true) (retail AnimDoneHook::Execute
0x00526c20 → Hook_AnimDone 0x0050fda0 chain), UseTime once per tick
(0x00517d57/0x00517d67); IMotionDoneSink bound to an
ACDREAM_DUMP_MOTION recorder (AD-36 — R3 rebinds to
MotionInterpreter.MotionDone).

Conformance: the 11-scenario pre-cutover trace suite (a6235a36) replays
green with six EXPECTED-DIFF annotations (double-hop walk-run routing,
reversed settle links, Branch-4 physics-only modifiers, Branch-1
style-change links, post-adjust mirrors) — everything unannotated is
byte-identical. Legacy suites repaired by a dedicated agent (fixtures
gain the retail-mandatory StyleDefaults + class-bit-tagged ids;
reflection state-poking replaced with real dispatch; zero category-D
regressions — the suspected cursor bug was a fixture class-bit gap),
plus two vacuously-passing tests fixed. Register: stale IA-4 deleted
(R1-P1 ported the negative-factor swap).

Full suite green: 3,458 passed (374+425+713+1946).

Closes H6, H7, H8, H9, H10-adapter, H16-wiring (r2-port-plan.md §3 Q4).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

docs/architecture/retail-divergence-register.md

@@ -37,14 +37,13 @@ accepted-divergence entries (#96, #49, #50).
 
 ---
 
-## 1. Intentional architecture (IA) — 17 rows
+## 1. Intentional architecture (IA) — 16 rows
 
 | # | Divergence | Where (file:line) | Why it is safe / justified | Risk if assumption breaks | Retail oracle |
 |---|---|---|---|---|---|
 | IA-1 | Contact-plane pre-seed on grounded movers (**#96 ACCEPTED** per ISSUES.md) — retail's `CTransition::init` clears `contact_plane_valid`; we seed from the body's previous-frame plane | `src/AcDream.Core/Physics/PhysicsEngine.cs:919` | Removing it broke last-step stair `step_up` (`892019b`, reverted); seed propagates the body's *real current* plane, behavior matched retail in the A6.P3 gates | A stale pre-seeded plane lets `AdjustOffset` project sub-step 1 onto a plane retail wouldn't have yet — wrong slope motion / step-up acceptance right after leaving a surface | `CTransition::init`, pc:272547 family |
 | IA-2 | Lateral self-heal beyond retail's keep-curr: when no candidate contains the sphere, try `FindVisibleChildCell` over the claim's stab-list before keeping the claim | `src/AcDream.Core/Physics/CellTransit.cs:912` | Reuses the recovery retail's own `AdjustPosition` performs (:280028 stab-list mode), applied at the `find_cell_list` site to heal near-miss claims without a doorway crossing | In containment-gap geometry, membership flips to a neighbouring room where retail keeps curr — wrong render root / collision cell at gap positions | `find_cell_list` keep-curr pc:308788-308825; `find_visible_child_cell` :311444 |
 | IA-3 | `get_state_velocity` prefers dat cycle velocity (`MotionData.Velocity × speedMod`) over the decompiled constant; constant kept only as max-speed clamp | `src/AcDream.Core/Physics/MotionInterpreter.cs:315` | Retail's constant equals the Humanoid RunForward `MotionData.Velocity`, so both paths agree on retail dats; dat is ground truth for other MotionTables (r03 §1.3) | Where dat velocity ≠ constant, body speed differs from the retail binary — DR / observer drift on exotic creatures or modded dats | `FUN_00528960`; `_DAT_007c96e0` RunAnimSpeed |
-| IA-4 | `MultiplyFramerate` omits retail's negative-factor StartFrame↔EndFrame swap (direction encoded in Framerate sign instead) | `src/AcDream.Core/Physics/AnimationSequencer.cs:129` | Our callers (ForwardSpeed updates) only pass positive factors; Advance loop handles negative framerates against StartFrame as lower bound | A future negative-factor caller (reverse playback) scales without swapping bounds — wrong frame range traversal instead of clean reversal | `FUN_005267E0`; ACE Sequence.cs L277-287 |
 | IA-5 | Per-ENTITY vertex-derived AABB culling (+5 m animated-drift margin; animated entities bypass cull) vs retail per-PART dat drawing spheres | `src/AcDream.App/Rendering/Wb/WbDrawDispatcher.cs:693` (bounds at `src/AcDream.Core/World/WorldEntity.cs:153`, `src/AcDream.Core/Meshing/GfxObjBounds.cs:14`; dead `PerEntityCullRadius=5.0f` at dispatcher :210) | Batched MDI rendering can't cheaply cull per part; bounds derive from the SAME dat vertex data that gets drawn (containment by construction — the **#119** fix, `6a9b529`; memory: feedback_culling_bounds_from_drawn_data) | Geometry escaping bounds+margin (pose drift >5 m, a hydration path skipping `SetLocalBounds`) makes the whole entity vanish on-screen — the #119 vanishing-staircase class | `CGfxObj.drawing_sphere` / viewconeCheck 0x005a09a4 |
 | IA-6 | Chat scrollback 500 lines vs retail ~200 (configurable) | `src/AcDream.Core/Chat/ChatLog.cs:19` | Strictly more useful for a dev client + plugins; deliberate default | Negligible — only if a plugin/UI behavior is ever specified against retail's exact retention cap | retail chat scrollback (~200) |
 | IA-7 | PhysicsScript replay keyed by (scriptId, entityId) replaces the prior instance; retail's ScriptManager linked list could hold duplicates | `src/AcDream.Core/Vfx/PhysicsScriptRunner.cs:51` | Prevents duplicate-stacking on server retriggers; flat keyed list simpler than retail's linked schedule; hedged to retail's common path | A server intentionally layering the same script on the same object shows ONE effect where retail shows several (overlapping casts/impacts) | `ScriptManager::Start` FUN_0051be40 / tick FUN_0051bfb0 |
@@ -61,7 +60,7 @@ accepted-divergence entries (#96, #49, #50).
 
 ---
 
-## 2. Adaptation (AD) — 34 rows
+## 2. Adaptation (AD) — 35 rows
 
 | # | Divergence | Where (file:line) | Why it is safe / justified | Risk if assumption breaks | Retail oracle |
 |---|---|---|---|---|---|
@@ -99,10 +98,11 @@ accepted-divergence entries (#96, #49, #50).
 | AD-33 | `CSequence.frame_number` at C# `double` (64-bit); retail is x87 `long double` (80-bit extended) — every frame-boundary comparison ran at extended precision on retail (Phase R1, 2026-07-02) | `src/AcDream.Core/Physics/Motion/CSequence.cs` (`FrameNumber`) | `double` is the widest C# float type; the R1 port removes ACE-style boundary epsilons so comparisons are exact-int against bare boundaries, minimizing ULP sensitivity (ACE's `float` is far worse) | A frame landing within 1 double-ULP of an integer boundary could classify differently than retail's 80-bit compare — sub-frame timing skew at pathological framerate×dt combinations | `acclient.h:30747` (`long double frame_number`) |
 | AD-34 | Retail's intrusive `DLListBase`/`DLListData` lists are managed `LinkedList<T>`s; node identity via `LinkedListNode<>` references (Phase R1 anim list, 2026-07-02; extended R2-Q3 to `MotionTableManager.pending_animations`) | `src/AcDream.Core/Physics/Motion/CSequence.cs` (`_animList`); `src/AcDream.Core/Physics/Motion/MotionTableManager.cs` (`_pendingAnimations`) | Same topology + cursor semantics (curr_anim/first_cyclic/tail-anchored scans are node references); unlink/delete becomes `Remove(node)`; conformance tests pin the surgery state tables | Any retail behavior depending on the −4 pointer adjustment or node memory reuse (none observed in the decomp) would diverge | `acclient.h` DLListBase; `r1-csequence-decomp.md` §0; `r2-motiontable-decomp.md` §11 |
 | AD-35 | `MotionTableManager.PerformMovement`'s unhandled-type default case returns the named sentinel `0xFFFFFFFF` (`MotionTableManagerError.NotHandled`); retail's compiled code leaks the `CSequence*` pointer reinterpreted as the return code (BN-confirmed artifact, dead/unreachable — callers gate on type first) (R2-Q3, 2026-07-02) | `src/AcDream.Core/Physics/Motion/MotionTableManager.cs` (`PerformMovement` default case) | No retail caller consults the return value for unhandled types (RawCommand/StopRawCommand/MoveTo\*/TurnTo\* route elsewhere); returning a stable non-zero sentinel preserves the only observable contract (non-zero = not success) without fabricating a pointer-shaped number | If a future port wires a caller that passes unhandled types AND branches on the exact return value, it would see `0xFFFFFFFF` where retail saw an arbitrary pointer — flag at that port | `PerformMovement` 0x0051c0b0 (`r2-motiontable-decomp.md` §11 default-case note) |
+| AD-36 | `IMotionDoneSink.MotionDone` is OBSERVED, not consumed: R2-Q4 binds the seam to an ACDREAM_DUMP_MOTION diagnostic recorder (`GameWindow` TickAnimations bind site); retail's chain continues `CPhysicsObj::MotionDone` → `MovementManager::MotionDone` → `CMotionInterp::MotionDone` popping `pending_motions` (R2-Q4, 2026-07-02) | `src/AcDream.Core/Physics/AnimationSequencer.cs` (`MotionDoneTarget`) | Stage boundary: `CMotionInterp.pending_motions` does not exist until R3-W2, which rebinds the seam to `MotionInterpreter.MotionDone` (r3-port-plan.md §4) | Until R3, nothing tracks queued-motion completion on the interp side — jump_is_allowed's pending-head peek and action RemoveAction mirrors are inert (they don't exist yet either) | `CPhysicsObj::MotionDone` 0x0050fdb0; r2-port-plan.md §4 |
 
 ---
 
-## 3. Documented approximation (AP) — 69 rows
+## 3. Documented approximation (AP) — 72 rows
 
 | # | Divergence | Where (file:line) | Why it is safe / justified | Risk if assumption breaks | Retail oracle |
 |---|---|---|---|---|---|
@@ -180,6 +180,8 @@ accepted-divergence entries (#96, #49, #50).
 ---
 
 | AP-73 | **Remote motion sink maps retail's substate+modifier animation model onto ONE cycle** — the L.2g S2 funnel dispatches retail-verbatim (style/forward/sidestep/turn per `apply_interpreted_movement` order + action-stamp gate), but `RemoteMotionSink.Commit` resolves the axes into a single `SetCycle` (forward locomotion > sidestep > turn priority + Run→Walk→Ready missing-cycle fallback); retail BLENDS modifiers over the substate cycle via `CMotionTable::re_modify` (DEV-9) | `src/AcDream.App/Rendering/RemoteMotionSink.cs` (`Commit`) | The pick + fallback are the pre-S2 behavior moved verbatim (user-verified across L.1/L.3); the STATE side is now fully retail through the funnel, so only the visible-animation composition approximates. Retire with the S3/S6 modifier port (`re_modify` + `subtract_motion` omega/velocity combine) | Run-while-turning shows pure run legs (no lean/turn blend); sidestep-while-walking shows one cycle only — animation-composition mismatch vs retail observers, position unaffected | `CMotionTable::re_modify` (GetObjectSequence tail, pc:298636); `StopSequenceMotion` modifier branch pc:298954 |
+| AP-74 | **K-fix18 `skipTransitionLink`** post-Q4 expression: `SetCycle(..., skipTransitionLink:true)` strips the just-dispatched transition links via `CSequence.RemoveAllLinkAnimations` so Falling engages instantly on jump start; retail gets the same strip from `LeaveGround` → `CPhysicsObj::RemoveLinkAnimations` (0x0050fe20) fired by the physics layer detecting ground departure (R2-Q4 rehost of the 2026-04-26 K-fix18) | `src/AcDream.Core/Physics/AnimationSequencer.cs` (`SetCycle` skip branch); GameWindow :4830/:10201 callers | Same primitive retail uses, fired from the caller instead of the ground-transition event; pending-queue tick counts intentionally not zeroed (retail's RemoveLinkAnimations doesn't touch the manager queue — the countdown chain absorbs the stall) | If a code path sets the flag outside a genuine ground-departure, links retail would play get eaten; queue completion for the stripped entry reports late (next sentinel) | `LeaveGround` 0x00528b00 → `RemoveLinkAnimations` 0x0050fe20; retire in R3-W4 (flag + call sites deleted, LeaveGround fires it) |
+| AP-75 | **Adapter-boundary `adjust_motion` + locomotion velocity/omega synthesis**: `SetCycle` still (a) remaps TurnLeft/SideStepLeft/WalkBackward to their mirrors with negated speed BEFORE dispatch (retail adjusts in `CMotionInterp` — R3 scope; GameWindow's local-player path passes raw ids), and (b) post-dispatch overwrites sequence velocity (low-bytes 05/06/07/0F/10) and omega (0D/0E, only when dat-silent) with the retail locomotion constants — retail drives BODY velocity from `get_state_velocity`, not the sequence accumulators (R2-Q4 carry-over of the pre-Q4 adapter tail) | `src/AcDream.Core/Physics/AnimationSequencer.cs` (`SetCycle` head remap + tail synthesis) | (a) preserves unadjusted GameWindow callers until R3-W6 unifies the local player onto MotionInterpreter; (b) preserves the remote-DR and local Option-B velocity consumers until R6 root motion drives the body (sibling of IA-3) | (a) none while callers stay in the known set; (b) a dat whose locomotion MotionData carries a REAL velocity different from the constants gets overwritten — exotic-creature speed skew (same class as IA-3's risk) | `CMotionInterp::adjust_motion` @305343; `get_state_velocity` 0x00528960; retire (a) R3-W6, (b) R6 |
 
 ## 4. Temporary stopgap (TS) — 33 rows
 
@@ -221,7 +223,7 @@ accepted-divergence entries (#96, #49, #50).
 
 ---
 
-## 5. Unclear (UN) — 4 rows
+## 5. Unclear (UN) — 5 rows
 
 These rows have a missing, contradictory, or never-argued justification.
 They are the highest-priority audits: each needs either a recorded