Fix pitch shift on mid-stream format change

[maximmaxim345]

When aiosendspin (as a server) changes sample rate via set_preferred_format, it sends a stream/start for the already-active role. Per the Sendspin spec, this updates the stream configuration without clearing buffers.

The worker immediately called player.set_format(), closing the old sounddevice stream and opening a new one at the new sample rate. Old PCM data still queued at the previous rate would then play through the new stream, causing pitch distortion.

With this PR, the worker now waits for the player to drain its internal queue before switching format. Incoming new-format chunks are buffered locally during the drain and submitted after the switch completes. Control items (stop, clear, volume) continue to be processed during the wait.

[Copilot]

Pull request overview

This PR addresses pitch distortion when the server changes the active stream’s sample rate mid-stream by avoiding a format switch while old-format PCM is still pending in the client’s playback pipeline.

Changes:

• Add AudioPlayer.is_drained() to detect when the internal playback queue is empty.
• Update the audio worker to wait for the player to drain before calling set_format().
• Buffer incoming chunks during the drain and submit them after the format switch, while still processing control messages (stop/clear/volume).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 5 comments.

File	Description
sendspin/audio_connector.py	Implements drain-before-switch logic and local buffering of chunks during the drain window.
sendspin/audio.py	Adds AudioPlayer.is_drained() to support safe mid-stream format switching.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.