fix(voice): use monotonic STT event deadlines

[hsusul]

Summary

• use time.monotonic() for OpenAI STT event timeout accounting
• update the timeout regression test to control the monotonic clock

Testing

• python -m compileall -q on changed source and test files
• focused pytest was attempted but this fresh worktree could not import the repository because optional dependency httpx2 was unavailable; dependency sync was blocked by package-index DNS failures

[seratch]

Can you fix the CI errors?

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: aad26e6b2e

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep the monotonic mock away from the event-loop clock

Patching process-global time.monotonic also replaces the clock used by asyncio.BaseEventLoop.time(). During this async test, event-loop scheduling consumes the five-value side_effect and can raise StopIteration, strand pending tasks, or expire asyncio.wait_for independently of the STT deadline, so the test is failing or passing for the wrong reason. Mock only an STT-local clock dependency, or otherwise preserve the loop's real clock.

AGENTS.md reference: AGENTS.md:L200-L200

Useful? React with 👍 / 👎.