test(sprites): hoist lockAllAnchors fixture onto a real-locked prototype

[Bryandero98]

Summary

• Addresses CI: cut fixture-construction file I/O in sprites/walk and sprites/atlas (successor to #6004) #6180's lockAllAnchors half: 136 call sites across walk.test.js/atlas.test.js/animationTrackWorkflow.test.js locked a character from byte-identical cached candidate PNGs, each re-running the real Sharp normalize + chroma-key-selection pipeline from scratch.
• lockAllAnchors now locks ONE real prototype per TEST_ROOT (at the full anchor set, since every caller's directions is a subset of it) and materializes every other character by recursive-copying only the artifacts a caller actually asked for, id-normalizing the manifest, and resetting any anchor outside the requested set to the exact seeded-pending shape a fresh partial lock would leave.
• No call site changed except the three that create the record + now pass records through — the fast path lives entirely inside spriteTestFixtures.js.
• Each suite gets an equivalence test (the acceptance gate CI: cut fixture-construction file I/O in sprites/walk and sprites/atlas (successor to #6004) #6180 asks for, not an extra): builds a character both through the real pipeline (lockAllAnchorsReal, exported unchanged) and through the materialized path, and asserts byte-identical results after id normalization — file names, PNG bytes, the full manifest (every embedded sha256), and the record's chromaKey/status.

Two options for this issue

This is option A of two — see #6207 for option B, which adds a buildFinalizedWalkSet optimization on top of this same lockAllAnchors change. I found that a copy-based materialization of buildFinalizedWalkSet (mirroring this PR's approach) doesn't actually help — that function never calls Sharp (writeWalkFramePng already caches the encoded buffer), so copying files isn't cheaper than the cached-buffer writes it already does. #6207 instead parallelizes that function's 8 independent per-direction writes, which is a real win with much less mechanism. I'd merge #6207 over this one — same lockAllAnchors change, plus a real additional win, at lower risk (no new copy/rewrite logic, so no new class of drift to worry about). This PR stands alone as the smaller, more conservative option if you'd rather not take the buildFinalizedWalkSet change.

Only #6207 carries "Closes #6180" — merging either is fine, but I didn't want both auto-closing the issue out from under the other.

Test plan

• Full server/services/sprites/ suite: 716 passed, 1 skipped, 0 failed (same pass count as main)
• Both suites' new equivalence tests pass, and caught two real bugs during development (south-anchor sync, partial-lock manifest status) before landing
• Measured locally (Windows): walk.test.js ~29.4s → ~19.7-21.4s tests phase (~30% faster); atlas.test.js ~24.4s → ~20.2-21.3s tests phase (~15% faster)
• Windows CI ⏱ before/after (will follow up with the actual CI numbers once this runs)

🤖 Generated with Claude Code

[Bryandero98]

Rebased onto main after #6205 merged (span narrowing + capSharpThreads) — both touch the same 3 files. No functional changes here beyond the merge; full sprites suite re-verified post-merge (26/26 files, 716/717 tests, 1 pre-existing skip).