feat(voice): apply Pocket speed changes during active speech

[johnmatthewtennant]

Summary

Pocket playback speed was previously captured when an utterance started, so changing the setting affected only later speech. This makes speed changes apply to the current preview, one-shot utterance, or streaming response without restarting playback. The native audio graph remains stable through 1x transitions, and spoken-delivery progress stays conservative and monotonic as the rate changes.

Reviewer-reproducible examples

1. Select Pocket for speech output and start a voice conversation with a multi-paragraph response.
2. While the response is speaking, change playback speed from 1x to 2x, then back to 1x.
3. Confirm the current utterance changes speed without restarting, dropping queued speech, or changing pitch.

[morgmart]

🤖 Automated code review

APPROVE. Fresh static review of the exact three-dot comparison found no publishable findings. Final self-check covered the changed live-speed flows for preview, one-shot speech, and streaming speech; accessibility and localization impact; navigation and consent guards; async errors, never-completes behavior, lifecycle and races; test honesty; project and design-system rules; duplicate overlap; and the evidence required for blocking severity. The supplied GitHub evidence reports all eight check runs completed successfully, while the legacy combined commit status is pending with no status contexts; required checks still independently govern merge readiness.

Deterministic publication result: 0 blocking and 0 non-blocking finding(s) publishable; 0 duplicate(s) suppressed.

Pending checks: 1 check(s) are not complete.

This approval reflects the completed code review only; merge readiness remains governed by the repository's required checks.