Bind queued speech receipts when their playback context is emitted

When a voice announcement is queued before caller-turn admission settles, do not permanently bind its completion receipt to the turn number at enqueue time. Bind it when its matching TTS playback context is emitted; otherwise an earlier caller response admitted late can make uninterrupted announcement audio appear stale. Keep invalidation for caller turns or interruptions after binding, and test both orders. In Pipecat, an AggregatedTextFrame with the assigned context can provide this boundary. This fixes an ownership race, not proof of handset audibility; downstream delivery still requires independent evidence.

Applies when: queued startup announcements overlap asynchronous admission of an earlier caller response. Keywords: Pipecat TTS startup playback receipt interruption turn admission race