Configure native audio source from device config instead of hardcoded defaults

[MaxHeimbrock]

Problem

When the device's audio output configuration differs from the hardcoded defaults — most commonly with a Bluetooth headset — microphone capture fails with sample_rate and num_channels don't match and the RtcAudioSource metadata-mismatch warning.

Root cause: RtcAudioSource created the native (Rust) audio source with a fixed sample_rate (48000) and num_channels (2). But captured frames flow through Unity's audio graph (AudioProbe.OnAudioFilterRead) at the actual DSP output configuration. The Rust native source does not resample — NativeAudioSource::capture_frame rejects any frame whose rate/channels differ from how the source was configured.

Change

• RtcAudioSource now reads the sample rate and channel count from Unity's output configuration (AudioSettings.GetConfiguration) instead of hardcoded defaults, falling back to the platform defaults only when Unity can't report one.
• The base constructor exposes two overloads: a device-mode one (RtcAudioSourceType only) for capture sources whose format is whatever Unity delivers, and an explicit one (type, sampleRate, channels) for sources that generate a fixed, known format.
• MicrophoneSource and BasicAudioSource use device mode; BasicAudioSource drops its unused channels parameter. SineWaveAudioSource (tests) declares its exact format.
• If a frame's format still doesn't match the source (an inconsistent Unity report, or the output configuration changing at runtime), the frame is dropped with a throttled warning rather than sent as a mismatch the native side would error on.
• Removes the redundant Microphone.Start(null, …, 44100) in the Meet sample.

Scope

This is the focused "get the format right at init" fix. It deliberately does not attempt to hot-swap the source mid-call when the audio device changes after publishing — that's a larger change (it requires recreating + republishing the track, which the binding between a track and its source handle forces) and is better handled separately, e.g. by having MicrophoneSource react to AudioSettings.OnAudioConfigurationChanged and restart capture. A device change during a call will currently drop frames (quiet audio) rather than auto-recover.

Verification

• Compiled the Runtime, PlayModeTests, and Meet Assembly-CSharp assemblies cleanly.
• Not yet run: live E2E PlayMode tests (require livekit-server --dev) and an on-device Bluetooth repro.

🤖 Generated with Claude Code