WIP: auto re-anchor a sendspin client stuck in error while playing - #351

You have commit messages with errors ⧗ input: wip(sendspin): auto re-anchor a client stuck in error while playing After Spotify single-active-device switching, a room's sendspin client can desync — the reference client logs "Audio underflow detected; requesting re-anchor" and re-anchors its own timeline, but when that client-local re-anchor does not converge the room plays nothing while ALSA stays RUNNING. Today the only recovery is restarting the sendspin@room_<x> client process. The protocol carries no inbound re-anchor request, so the sole server-visible signal is a `state: 'error'` player-state report (identical repeats are deduped in handleClientState, so we act on the first transition, debounced). If the client is still in `error` after a short debounce while the output believes it is playing, force a fresh server anchor (startStream preserveAnchor:false) — what a client restart does, without the restart — guarded by a per-output cooldown. WIP / unverified: needs a live capture to confirm (a) the reference client reports `state: 'error'` during this underflow storm and (b) a server-side fresh anchor converges it rather than needing a full client restart. Opening for direction. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> ✖ type must be one of [build, chore, ci, docs, feat, fix, perf, refactor, revert, style, test] [type-enum] ✖ found 1 problems, 0 warnings ⓘ Get help: https://github.com/conventional-changelog/commitlint/#what-is-commitlint

Node.js 20 is deprecated. The following actions target Node.js 20 but are being forced to run on Node.js 24: actions/checkout@v4. For more information see: https://github.blog/changelog/2025-09-19-deprecation-of-node-20-on-github-actions-runners/