MacOS specific AEC implementation

[yujonglee]

#815

[yujonglee]

1. https://github.com/yury/cidre/blob/89211ec51ef0c2052baa0a5669c3b14ec3d4ca1f/cidre/examples/au-capture/main.rs#L67
2. https://github.com/yury/cidre/blob/89211ec51ef0c2052baa0a5669c3b14ec3d4ca1f/cidre/examples/core-audio-record/main.rs#L17

We currently use 2 for audio output capture. For 1, we use Cpal for mic but want to use apple's voice processing.

Current problem is that running 1/2 seperatly works, but if I run them together, we got empty audio in the audio-tap(system audio capture) callback. Seems like audio format mismatch or something? Not sure.

[yujonglee]

https://github.com/toeverything/AFFiNE/blob/canary/packages/frontend/native/media_capture/src/macos/tap_audio.rs

Affine just route mic+tap to single aggregate device. We want to be split mic/speaker to be flexible, but maybe it is impossible or not worth it?
(Affine also listens to device change to ensure stay-up-to-date audio format)