fix(player): pause parent audio proxy on seek to prevent stutter loop

[terencecho]

Summary

• seek() only called seekAll() under parent audio ownership, leaving the <audio> proxy playing while the timeline froze at the new seek target.
• The periodic mirrorTime drift-correction (parent-media.ts) would then yank currentTime back to the timeline position every ~80ms of accumulated drift, producing an audible stutter loop while the video frame stayed frozen.
• Fix: make seek() symmetric with pause() — pause the parent proxy before seeking it.

Repro

1. Use the player in an environment where the runtime posts media-autoplay-blocked (mobile / autoplay-restricted contexts), promoting audio ownership to "parent".
2. Start playback with audio.
3. Click anywhere on the scrubber while playing.
4. Before: audio stutters in a short loop while the video frame is frozen.
5. After: audio cleanly pauses at the new seek position.

Test plan

• Added regression test `seek() while playing pauses parent proxy (prevents mirrorTime stutter loop)` in `hyperframes-player.test.ts`.
• `pnpm --filter @hyperframes/player test` — 110/110 pass.
• Manual repro on a device where ownership flips to `parent`.

🤖 Generated with Claude Code

[jrusso1020]

LGTM — clean symmetric-with-pause fix for the mirrorTime stutter loop.

Audited

The root-cause framing is precise: seek() set this._paused = true and called seekAll(target), but never called pauseAll() on the parent audio proxy. So the proxy kept playing past the now-frozen timeline target, and the periodic mirrorTime drift-correction (parent-media.ts) yanked currentTime back every ~80ms of accumulated drift → audible stutter loop while the video frame stayed frozen.

The fix is exactly the missing call:

if (this._media.audioOwner === "parent") {
  this._media.pauseAll();      // ← added
  this._media.seekAll(timeInSeconds);
}

The inline comment captures the failure mode in one sentence: "leaving the proxy playing turns the next mirrorTime drift-correction tick into a perpetual seek→play→drift→seek stutter loop." That's exactly the right thing to pin for future readers. ✓

The fix is scoped to audioOwner === "parent" — the runtime-owned path isn't touched. That's correct because runtime-owned audio lives inside the iframe and is driven by the runtime's own seek-handling logic; the parent-proxy path is specifically for autoplay-blocked / mobile contexts where audio plays from the parent's <audio> element. The bug only manifests under that ownership, and the fix is scoped to it.

Test coverage

The new regression test is right-shaped:

player._promoteToParentProxy?.();
player.play();
mockAudio.pause.mockClear();   // isolate from play()'s incidental work
player.seek(12.5);
expect(mockAudio.pause).toHaveBeenCalled();
expect(mockAudio.currentTime).toBe(12.5);

Pins both the pause call AND the seek target. The mockClear() between play and seek isolates the assertion from any pause calls play() might have made. ✓

Body claim verification

• "seek() only called seekAll() under parent audio ownership, leaving the proxy playing" — verified by the diff ✓
• "mirrorTime drift-correction (parent-media.ts) would then yank currentTime back to the timeline position every ~80ms" — citation to out-of-diff code; plausible mechanism
• "Fix: make seek() symmetric with pause() — pause the parent proxy before seeking it" — verified ✓

Non-blocking note

Path-enum for other seekAll-without-pauseAll call sites: the bug pattern is "set audio proxy currentTime while the parent's mirrorTime drift-correction is still running." Worth a quick grep for other entry points that could trigger the same shape — frameStep, stepFrame, programmatic currentTime setters, or any other public method that mutates timeline state under audioOwner === "parent". If there are other callers of seekAll (or direct audio.currentTime = ... writes), they may have the same drift-loop bug at a different entry point. The pause-before-seek invariant should be centralized — either every seekAll call site does the pause defensively (current pattern), or seekAll itself calls pauseAll first (cleaner contract). Cosmetic for this PR; worth a follow-up audit.

CI

Preflight, Detect changes, Format, Semantic PR title, File size check, regression all green so far. Test, Typecheck, Lint, Build, CLI smoke (required), Perf:*, Analyze (*) in-progress on this commit. No failures.

Test plan checkbox "Manual repro on a device where ownership flips to parent" still unchecked — the regression test pins the unit behavior; the manual integration verification on a mobile/autoplay-blocked device is the remaining gate. Worth doing once before merge but not blocking.

— Review by Rames Jusso (pr-review)