[T5.2] Tiles wall stops re-animating on the cut (persistent-sources slice 2, on T5.1)

[iamfatness]

Backlog item T5.2 (tier 5) in docs/BACKLOG.md.

Tiles wall stops re-animating on the cut (persistent-sources slice 2, on T5.1)

Clause / size / why: 3 | M-L

Order lives in BACKLOG.md; status lives here.

[iamfatness]

PR #511 implements this — slice 2 of the persistent-sources redesign, and the first real consumer of the #419 SourceRegistry.

The root cause turned out to be one level deeper than the first fix assumed. TilesPlanAnimation::adoptSettledFrom (the 2026-09-09 hand-over) could only move a settled wall, because with two per-bus animators mid-flight state had no correct owner — so a wall taken mid-animation still broke. PR #511 replaces the hand-over with one animator per wall (core::TilesWallSource, keyed by wall id, shared by both buses): there is nothing to hand over, and the cut is continuous even mid-flight. adoptSettledFrom is deleted.

Correction to this issue's own description of the symptom: the wall does not "replay from alpha 0". TilesAnimator treats a reset animator's next non-empty sample() as an adoption, so the wall snaps to its final state — alpha pops to 1, mid-spring rects jump to settled. That is the opposite direction, and it matters: an EXPECT_GE(alpha_after, alpha_before) assertion is satisfied by a snap just as well as by continuity and catches nothing (the first draft of the headline test passed against the unfixed code).

Verified: 978 native tests green on a Release core; SHOW DRILL PASSED at 8×1080p60 with 100% delivery; and live-meeting-soak.mjs --minutes 5 --takes 20 --sources 8 → 20 takes, 20 cut, 0 rebuilt over a shared 8-member wall, 60.0/s render with 0 deadline misses, 0 dropped video and 0 underruns.

Remaining for plan 2 (the wall-texture slice): the one-layer compositor, Metal + CPU-preview parity, the multiview PVW cell as the registry's first real reader, and AtomicTakeCoordinator. Residual minors are filed as #512.