fix(physics): #265 landing-bounce family - retail check_contact seed + velocity-free landing commit

Retail jump landings BOUNCE: the floor touch records both a contact plane
(grounding) AND a collision normal (collided_with_environment), and
handle_all_collisions reflects the unmodified impact velocity off it at
5% elasticity (v += -(v.n)(elasticity+1).n, DEFAULT_ELASTICITY 0.05
@0x007c6a7c). Our transition already recorded both facts; the bounce was
suppressed by the AD-25 adaptation stack in the per-tick commit: a
Velocity.Z<=0 landing gate (needed because the resolver glued ascending
movers to the ground) plus a landing Velocity.Z=0 hand-zero whose stated
purpose was making the reflect a no-op. Downhill glided instead of
bouncing, flat-ground landings had no pop, and uphill jumps flapped
between grounded/airborne against the animation machine.

Three retail mechanisms replace the stack:
- check_contact (0x0050f5b0) seeding in ResolveWithTransition: a body in
  CONTACT seeds the transition's contact only while v.contactPlane.N <=
  0.0002; moving away seeds the last-known plane alone (get_object_info
  0x00511cc0). Ascending jumps therefore run contact-free (ballistic, no
  glue) - the gate's reason-for-being is gone. The plane requirement is
  strict: Contact-without-plane is unrepresentable in retail.
- SetPositionInternal-shaped commit (0x00515330, byte-read end-to-end,
  velocity-sign-FREE): contact purely from the transition's contact
  plane, HitGround on the airborne->walkable edge, HandleAllCollisions
  with unmodified impact velocity. Whole commit gated on Ok &&
  candidateMoved (retail pc:283657 skips SetPositionInternal entirely
  when the candidate did not move) - a standing body's contact state is
  never re-derived, which is what keeps rest bit-stable (AD-41 updated).
- Byte decodes: gate override state&0x800000=Sledding, zero branch
  state&0x20000=Inelastic, reflect strictly dot<0 - our port already had
  all three correct.

Settle: real landings (>=0.25 m/s) bounce and decay geometrically;
smaller impacts are consumed by retail's unconditional small-velocity
zero, so standing never micro-bounces. Re-baselines documented in place:
landing-survival pin measures decay post-settle; LiveCompare_Tick0/376
pin the new IsOnGround=false on zero-move ticks (captured true was the
retired seed echo; tick 376's captured body carries an 11.8 m/s grounded
velocity from the deleted get_state_velocity-overwrite era); de-overlap
fixture now carries the plane real grounded bodies always have. New
pins: LandingBounceSeedingTests (ascent no-seed, rest keeps contact,
strict plane, slope 5% reversal + tangential preservation, Sledding
override).

Investigation + implementation record:
docs/research/2026-07-30-landing-bounce-family.md. Complete Release
suite: 10,031 passed / 5 skips / 0 failures.

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

docs/architecture/retail-divergence-register.md

@@ -108,7 +108,7 @@ accepted-divergence entries (#96, #49, #50).
 | AD-37 | Camera rotation state is a forward VECTOR (nlerp + normalize; roll always 0, up = world Z); retail's sought carries a full Frame and slerps quaternions (`Frame::interpolate_rotation` shortest-path slerp with 2e-4 nlerp fallback). The dead-band compares forward-vector distance against the same 2e-4 epsilon retail applies per quaternion component | `src/AcDream.App/Rendering/RetailChaseCamera.cs` (`_dampedForward`, `ApplyConvergenceSnap`) | The chase camera never rolls (heading frames are Z-up by construction), so a forward vector spans the reachable rotation space; identified (not introduced) during the #180 UpdateCamera tail reading | If a future camera mode needs roll (death cam, cutscene) the vector state can't represent it; large-angle per-frame turns nlerp (chord) vs slerp (arc) — imperceptible at 0.45-stiffness step sizes | `Frame::interpolate_rotation` 0x00535390, `Frame::close_rotation` 0x00455d70; pseudocode doc 2026-07-06-camera-sought-position |
 | AD-39 | The `frames_stationary_fall` ladder + fsf≥3 UP-contact-plane manufacture runs AFTER acdream's fused LKCP-restore/contact-marking block, deriving retail's `_redo` as `cleanAdvance \|\| OnWalkable`; retail (ACE Transition.cs:1029-1061) interleaves the fsf block BETWEEN the LKCP-restore (sets `_redo`) and the contact-marking (reads the manufactured plane) (#182 rebuild, 2026-07-07) | `src/AcDream.Core/Physics/TransitionTypes.cs` (`ValidateTransition` fsf tail) | acdream deliberately fused ACE's separate LKCP-restore + contact-mark blocks (the L.2.3c/L.2.4/A6.P3 contact-retention divergences); running the ladder after them and re-marking grounding inside the manufacture branch is semantically equal (a grounded wall-slide is not a stuck-fall in either arrangement) without disturbing those hard-won fixes | If a future contact-retention change alters when OnWalkable is set relative to the ladder, `_redo` could misclassify a frame (grounded-jam mistaken for stuck-fall → spurious velocity zero, or vice-versa) — the fsf conformance tests pin the current arrangement | `CTransition::validate_transition` 0x0050aa70 pc:272625-656; ACE Transition.cs:1029-1061 |
 | AD-40 | The fsf `Stationary*` transient-bit encode (fsf→0x10/0x20/0x40) lives in the Core resolve writeback (`PhysicsEngine.ResolveWithTransition`), co-located with the fsf computation; retail encodes it in `handle_all_collisions` (pc:282737-758). Also: `PhysicsBody.CachedVelocity` is computed at the player chokepoint but not yet consumed — outbound wire velocity still uses the existing `get_state_velocity` path, not retail's cached_velocity source (#182 rebuild, 2026-07-07) | `src/AcDream.Core/Physics/PhysicsEngine.cs` (writeback); `src/AcDream.Runtime/Gameplay/PlayerMovementController.cs` (`CachedVelocity`) | Encoding in the writeback keeps the seed→ladder→writeback→seed round-trip self-contained in Core (testable without the App loop); the bit values + timing are identical to retail's (set after fsf is final, before the next resolve). CachedVelocity is faithful to carry now; routing the wire through it is a separate, unmeasured change | If a future consumer reads the Stationary* bits expecting retail's handle_all_collisions to have set them (it doesn't run in Core), the Core writeback is the source of truth; a wire-reporting change that assumes CachedVelocity is live would send the wrong velocity until it's wired | `handle_all_collisions` bit encode pc:282737-758; `get_velocity` 0x005113c0 (cached_velocity reader) |
-| AD-41 | The `candidateMoved` gate (retail UpdateObjectInternal pc:283657 `candidate != m_position`) suppresses ONLY `handle_all_collisions` + `cached_velocity` on a no-move frame; acdream still runs `ResolveWithTransition` (zero-distance) for cell/contact tracking, where retail skips the whole transition (#182 rebuild, 2026-07-07) | `src/AcDream.Runtime/Gameplay/PlayerMovementController.cs` (`candidateMoved` guard) | The load-bearing effect is not re-zeroing the gravity velocity that rebuilds after a stuck-fall bleed; the zero-distance resolve is a near-no-op (numSteps 0 → the zero-step early return, no ValidateTransition, contact plane persists via the writeback), so running it is harmless while keeping acdream's per-frame cell/membership refresh | If the zero-distance resolve ever gains a side effect on a no-move frame (a contact-plane clear, an fsf change), it would diverge from retail's skip — a no-move frame must stay a near-no-op | `CPhysicsObj::UpdateObjectInternal` 0x005156b0 pc:283657 (candidate-moved gate) |
+| AD-41 | The `candidateMoved` gate (retail UpdateObjectInternal pc:283657 `candidate != m_position`) suppresses the WHOLE SetPositionInternal-shaped commit (contact/walkable flags, HitGround/LeaveGround, `handle_all_collisions`, `cached_velocity`) on a no-move frame — narrowed 2026-07-30 (#265 bounce rework) from "only handle_all_collisions"; acdream still runs `ResolveWithTransition` (zero-distance) for cell/contact tracking, where retail skips the whole transition (#182 rebuild, 2026-07-07) | `src/AcDream.Runtime/Gameplay/PlayerMovementController.cs` (`candidateMoved` guard) | The load-bearing effect is not re-zeroing the gravity velocity that rebuilds after a stuck-fall bleed; the zero-distance resolve is a near-no-op (numSteps 0 → the zero-step early return, no ValidateTransition, contact plane persists via the writeback), so running it is harmless while keeping acdream's per-frame cell/membership refresh | If the zero-distance resolve ever gains a side effect on a no-move frame (a contact-plane clear, an fsf change), it would diverge from retail's skip — a no-move frame must stay a near-no-op | `CPhysicsObj::UpdateObjectInternal` 0x005156b0 pc:283657 (candidate-moved gate) |
 | AD-42 | Enter-world placement is split across two Core calls: legacy `Resolve` performs retail `AdjustPosition` + the host's established floor snap, then `ResolvePlacement` runs the verbatim object-aware `find_placement_pos` ring search. Retail runs initial environment placement, ring search, and final step-down inside one `find_placement_position` transition | `src/AcDream.App/Rendering/GameWindow.cs` (`EnterPlayerModeNow`); `src/AcDream.Core/Physics/PhysicsEngine.cs` (`ResolvePlacement`) | The first call has already committed the same validated cell/floor point that feeds the ring search; the second call uses the same sphere dimensions, collision registry, and cell id. Keeping the split preserves the proven indoor-login snap while adding the missing occupied-position behavior | A spawn that requires retail's final placement step-down after a ring candidate (rather than the existing floor snap before it) could settle at a slightly different Z on a ledge/water boundary; the overlap is still cleared | `CPhysicsObj::enter_world` 0x00516170; `CTransition::find_placement_position` 0x0050C170; `CTransition::find_placement_pos` 0x0050BA50 |
 | AD-43 | A malformed/custom PhysicsScript `CallPES` cycle whose script timeline never advances is rejected with a diagnostic; retail's linked scheduler would continue draining that zero-time tail indefinitely | `src/AcDream.Core/Vfx/PhysicsScriptRunner.cs` (timeline-progress ancestry guard) | Prevents corrupt DAT content from hanging the single update/render thread. Installed-DAT audit plus conformance tests prove the real rolling-weather cycles advance 2.8 seconds per edge and continue unchanged; only a no-progress strongly connected cycle is rejected | A custom DAT that deliberately relies on an infinite zero-time loop observes a rejected play instead of freezing the client | `ScriptManager::AddScriptInternal` 0x0051B310; `ScriptManager::UpdateScripts` 0x0051B480; `CPhysicsObj::CallPES` 0x00511AF0 |
 | AD-44 | acdream has no retained character-management screen: startup deterministically selects the first active, non-greyed CharacterList identity, and native-window close performs retail's complete character-logoff handshake plus transport disconnect before exiting instead of returning to character selection. One active `ReceiverData` equivalent means `ClientNet::LogOffServer`'s per-receiver loop sends one header. | `src/AcDream.Core.Net/Messages/CharacterList.cs` (`TrySelectFirstAvailable`); `src/AcDream.App/Rendering/GameWindow.cs` (live-session bootstrap, moving to `LiveSessionController` in Slice 3); `src/AcDream.Core.Net/WorldSession.cs` (`SelectCharacterForEnterWorld`, `Dispose`); `src/AcDream.Core.Net/Packets/TransportDisconnect.cs` | This preserves unattended startup and immediate ACE endpoint release while validating that the chosen identity is active/non-greyed and using the server's canonical account. A future retained character-management owner is separate UI/session work. | An account with multiple playable characters enters the first wire-order identity without retail's explicit choice. An eventual in-client "log off character" action cannot reuse the process-exit path; it must retain the authenticated socket after server `0xF653` and return to character management. | `gmCharacterManagementUI::SelectCharacter @ 0x004EC160`; `gmCharacterManagementUI::EnterGame @ 0x004ED440`; `gmCharGenMainUI::Update @ 0x004E8460`; `Proto_UI::LogOffCharacter @ 0x00546A20`; `CPlayerSystem::RequestLogOff @ 0x00562DD0`; `CPlayerSystem::ExecuteLogOff @ 0x0055D780`; `ClientNet::LogOffServer @ 0x00543EF0`; `SharedNet::SendOptionalHeader @ 0x00543160` |