perf: 0.2.0 re-measured at 1440p — 379 fps average, 6.3x the floor

-qaselftest at 2560x1440, TSR on, VSync off, no frame generation: 759 frames,
avg 2.64 ms (379 fps), p50 2.54, p95 3.39, p99 3.76, max 4.05 ms. The body, skin
material and animation add about 2 % over 0.1.5's 2.61 ms.

The same run passed every QA check: walk, reset 0 cm from spawn, pause/unpause,
transect 1456 samples with 0 holes, wall probe 0/35, facade probe 0/9.

Series committed as Docs/perf/0.2.0-walk-frametimes.csv and the result recorded in
the release-checks table next to the 0.1.0 and 0.1.5 numbers.

TODO.md

@@ -34,7 +34,7 @@ Actionable task list for this project. This file is the single, central place fo
 
 - [x] Third-person traversal works end-to-end in the demo area. **Walking is captured and verified 2026-11-09**: `demo-third-person-walk` shows the body mid-stride in profile (legs apart, arms counter-swinging), and the run logs `walk=1 speed=28 clip=GigLine_Walk`. Getting there needed two harness fixes, both found by probing rather than guessing: the camera must track the pawn (`ShotFollow`), and the call that enables walking had to move out of the sequencer's `BeginPlay` — it ran before the player pawn existed, so the cast silently found nothing and `speed` stayed 0 while the probe reported `clip=Idle` every frame.
 - [ ] No blocking animation or rig defects.
-- [ ] Performance target still met at 1440p.
+- [x] Performance target still met at 1440p. Measured 2026-11-09 at **2560×1440, no frame generation, VSync off**: **759 frames, avg 2.64 ms (379 fps), p50 2.54, p95 3.39, p99 3.76, max 4.05 ms** — 6.3× the 60 fps floor. Body, skin material and animation cost about 2 % against `0.1.5`'s 2.61 ms. Series: `Docs/perf/0.2.0-walk-frametimes.csv`. The same run passed every QA check: walk, reset 0 cm, pause/unpause, transect 1456 samples / 0 holes, wall probe 0/35, façade probe 0/9.
 - [ ] Tag `v0.2.0` and write `CHANGELOG.md` entry; then remove this section.
 
 ## 0.2.1 — Rig fixes (patch)