feat(physics): R5-V2 — wire TargetManager voyeur system per-entity, retire AP-79

Replace the AP-79 P4 TargetTracker poll adapter with the ported retail
TargetManager voyeur subscription system, wired per entity. Behaviorally a
faithful no-op refactor for the common cases (server-directed creature chase
+ auto-walk-to-object), now driven by the retail mechanism instead of the
GameWindow poll.

New: EntityPhysicsHost (App) — the per-entity IPhysicsObjHost (retail
CPhysicsObj stand-in), owning a TargetManager. Delegate-injected accessors so
it stays free of GameWindow internals (code-structure rule #1).

Wiring (GameWindow):
- _physicsHosts registry (guid → host) = retail CObjectMaint::GetObjectA,
  backing the voyeur round-trip's cross-entity delivery.
- ResolvePhysicsHost lazily creates a minimal position-only host for ANY known
  entity — retail lets every CPhysicsObj host a target_manager, so a STATIC
  object (chest/corpse) still answers add_voyeur; without this, auto-walk to a
  never-animated object would arm the moveto but never receive the immediate
  target snapshot and never start.
- MoveToManager set_target/clear_target/quantum seams repointed at the host's
  TargetManager (remote + player).
- HandleTargetting ticked unconditionally per entity BEFORE UseTime (retail
  UpdateObjectInternal order): per-remote loop for remotes, pre-Update block for
  the player. The player's tick is load-bearing for creature-chase — the player
  as a watched target pushes its position to the chasing NPCs' HandleUpdateTarget
  each frame, ahead of their UseTime.
- Despawn (RemoveLiveEntityByServerGuid): NotifyVoyeurOfEvent(ExitWorld) to the
  entity's watchers before pruning the host — the only cleanup for a watcher
  whose target already sent an Ok (past Undefined, the 10s staleness never fires).

Deleted: RemoteMotion.TrackedTarget* fields + _playerMoveToTarget* GameWindow
fields + the two manual poll blocks. AP-79 register row retired same commit
(AP section 73→72).

Delivery is now synchronous-on-set_target (retail: set_target is last in
MoveToObject, so the immediate snapshot lands with all moveto state in place)
vs AP-79's next-frame poll — more faithful. Full suite 4006 green.

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

docs/architecture/retail-divergence-register.md

@@ -99,7 +99,7 @@ accepted-divergence entries (#96, #49, #50).
 
 ---
 
-## 3. Documented approximation (AP) — 73 rows
+## 3. Documented approximation (AP) — 72 rows (AP-79 retired R5-V2 — the P4 TargetTracker adapter replaced by the ported TargetManager voyeur system)
 
 | # | Divergence | Where (file:line) | Why it is safe / justified | Risk if assumption breaks | Retail oracle |
 |---|---|---|---|---|---|
@@ -177,7 +177,6 @@ accepted-divergence entries (#96, #49, #50).
 | 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 |
 | AP-76 | **Remote rotation from the ObservedOmega side-channel**: the R2-Q5 sink callbacks (`MotionTableDispatchSink.TurnApplied/TurnStopped`) seed `RemoteMotion.ObservedOmega = (0,0,-(pi/2)*signedTurnSpeed)` on wire turns and zero it on turn-stops; GameWindow's per-tick step applies ObservedOmega (preferring it over sequence omega) to the remote body's orientation. Retail rotates the body from the SEQUENCE omega inside the per-tick apply_physics chain (CSequence::apply_physics -> CPhysicsObj omega integration) (carried verbatim from the deleted RemoteMotionSink; H17) | `src/AcDream.App/Rendering/GameWindow.cs` (sink callbacks in OnLiveMotionUpdated + the omegaToApply step in TickAnimations) | Same angular rate retail derives (pi/2 rad/s x turn speed); starts rotation the same tick as the wire turn without waiting for R6's per-tick order; UpdatePosition orientation snaps bound any drift | If the dat's turn modifier omega differs from the pi/2 constant, remote rotation rate diverges until an orientation snap; double-application risk if R6 lands apply_physics-driven rotation without deleting this seam | retire in R6 (retail per-tick order: apply_physics drives remote rotation) |
 | AP-77 | **`apply_current_movement`'s interpreted-branch tail (`ApplyCurrentMovementInterpreted`) writes body velocity DIRECTLY via `get_state_velocity`/`set_local_velocity` when grounded, instead of dispatching through an `IInterpretedMotionSink`** — retail's real `apply_interpreted_movement` (0x00528600) drives velocity indirectly through `DoInterpretedMotion`'s animation-table backend (the SAME function the funnel's `ApplyInterpretedMovement` already uses WITH a sink); this dual-dispatch call site (`HitGround`/`LeaveGround`/`ReportExhaustion`/hold-key toggles/`SetWeenieObject`/`SetPhysicsObject`) has no sink threaded through it | `src/AcDream.Core/Physics/MotionInterpreter.cs` (`ApplyCurrentMovementInterpreted`) | Direct continuation of the pre-R3-W4 `apply_current_movement` approximation (R3-W4 plan explicitly keeps this shape rather than relocating it — "the direct grounded-velocity write MOVES to the controller-side call site unchanged"); correct for the grounded, no-animation-table-yet state acdream is in today | A MotionTable whose locomotion cycle bakes a DIFFERENT velocity than `get_state_velocity`'s constants would silently diverge from what the animation actually plays, since this path never touches the animation backend at all | `CMotionInterp::apply_interpreted_movement` 0x00528600; retire when a sink is threaded through `apply_current_movement`'s callers (R6 root motion) |
-| AP-79 | **P4 TargetTracker minimal adapter**: MoveToManager's setTarget/clearTarget seams store the tracked guid on RemoteMotion (remotes, R4-V4) or the `_playerMoveToTarget*` GameWindow fields (local player, R4-V5); GameWindow's per-tick block (remotes) / pre-Update feed (player) delivers `HandleUpdateTarget(Ok)` from the live entity table whenever the target moved beyond the set_target radius (and `ExitWorld` when the entity despawns) — retail's TargetManager is a full subscription system with per-target quantums and callback scheduling (R4-V4 + R4-V5, 2026-07-03) | `src/AcDream.App/Rendering/GameWindow.cs` (the tracker block in TickAnimations + the setTarget seam in EnsureRemoteMotionBindings; the player twin: `_playerMoveToTarget*` fields, the EnterPlayerModeNow setTarget seam, and the pre-Update feed in OnUpdateFrame) | The manager's deferred-start/retarget lifecycle only needs position deltas at the tracked radius; the adapter delivers exactly that from data the client already holds | Update cadence is frame-quantized rather than quantum-scheduled; a target moving sub-radius never re-fires (retail same); multi-listener semantics absent until R5's TargetManager port | `TargetManager::ReceiveUpdate` chain (r4-moveto-decomp.md); retire in R5 |
 | AP-80 | **PlanFromVelocity survives for velocity-only NPC cycles** (M16): UpdatePosition-derived speed picks Ready/Walk/Run cycles for server-controlled creatures whose UMs never arrive (scripted-path NPCs); retail derives every cycle from motion messages through the motion tables (R4-V4 note; pre-existing mechanism, row added per the V4 plan) | `src/AcDream.Core/Physics/ServerControlledLocomotion.cs` (`PlanFromVelocity`); consumer `GameWindow.ApplyServerControlledVelocityCycle` | Some ACE entities move by position updates alone — without this, they slide in T-pose; constants (StopSpeed 0.2, RunThreshold 1.25) tuned against live ACE traffic | Cycle-pick thresholds are acdream inventions — a creature intended to walk fast may show run legs near the threshold | retire in R6 (root motion + full per-tick order) |
 | AP-81 | **Remote-DR gravity toggled via the Gravity STATE bit**: the jump handler sets `Body.State \|= Gravity` at VectorUpdate and both landing blocks clear it after `HitGround()`; retail keeps GRAVITY set for the object's whole life and gates gravity ACCELERATION on the Contact transient (`calc_acceleration`) (pre-existing K-fix9/K-fix15 mechanism, row added during #161 — which also fixed the ordering so `Motion.HitGround()`'s verbatim `state&0x400` gate runs BEFORE the clear) | `src/AcDream.App/Rendering/GameWindow.cs` (VectorUpdate jump handler + the two landing blocks) | The DR tick integrates gravity only for airborne remotes; the flag dance delivers exactly that without porting the full contact-gated `calc_acceleration` chain; the #161 ordering fix keeps the retail HitGround contract satisfied | Any NEW call into `Motion.HitGround`/`LeaveGround` placed after the clear silently no-ops on the gravity gate (the #161 leg-2 class); grounded remotes carry a non-retail state word (probes comparing state bits vs retail mislead) | `CPhysicsObj::calc_acceleration` (contact-gated); `set_on_walkable` 0x00511310; retire in R6 (contact-gated accel + persistent GRAVITY) |