Generalize audio device selection to preflight checks (input selection & non-Meet output routing)

Summary

Currently, media preflight verifies microphone activity via passive volume threshold checks (audio-check.js) and audio output via subjective confirmation tone, without allowing candidates to pick specific hardware devices. While audio-output.js already implements robust output routing via setSinkId and device enumeration, it is strictly bound to Google Meet tab-presentation mode (meetOutputSelect).

I propose adding a microphone (audioinput) selector to the preflight gate and making audio output selection available for standard interview sessions.

Current Implementation Analysis

Audio Input:
- audio-check.js checks micPeak >= MIC_SILENT_PEAK against pool.trackOf("audio").
- Candidates with multiple microphones (e.g., built-in laptop mic vs. external USB condenser / headset) have no way to choose which device feeds into the preflight check or LiveKit room without altering OS-level defaults.
Audio Output:
- audio-output.js already provides serialized setSinkId routing and device enumeration (refreshAudioOutputs, applyAudioOutput).
- However, this is only rendered under nodes.meetOutputRow when presenting in Google Meet, leaving standard sessions unable to route Jim's voice explicitly to headphones.

Motivation

Built-in microphones often capture loud mechanical keyboard clicking during the REACTO coding phase, reducing Jim's speech recognition accuracy.
Multi-monitor setups often set HDMI/DisplayPort monitor speakers as the OS default, causing echo unless the user manually alters system settings before opening CodeTrial.

Proposed Changes

Preflight Microphone Selector:
- Query navigator.mediaDevices.enumerateDevices() for kind === 'audioinput'.
- Expose a <select> dropdown in the preflight gate to pass a specific deviceId constraint to the media track pool.
Expose Output Selection in Preflight:
- Decouple the device selection logic in audio-output.js from meet-audio.js / nodes.meetOutputRow so that candidates can route Jim's audio to their desired output device directly from the preflight screen.

References

web/audio-output.js: refreshAudioOutputs(), applyAudioOutput()
web/audio-check.js: preflightReadiness(), mediaReadiness()

@s4512138

@s4512138 Accepted as scoped: a microphone selector in the preflight, and the output selector from web/audio-output.js offered outside Meet mode as well. Keep the default device as the default, so a candidate who picks nothing sees no change. #44 and #197 (a silent test tone on some drivers) may be helped by choosing the output explicitly; note in the pull request whether you could check that.

I am working on this change following the agreed scope.

@s4512138

@s4512138 On 2026-10-12 you said you were working on the scope accepted on 2026-10-05, and no pull request has appeared since. Is it still coming? If there is no reply by 2026-10-27, the assignment is removed and the issue stays open for anyone else to take, with Closes #206.

@jserv

@jserv Yes I am still working on the change, thank you.