Add a sound recorder that writes several devices to one Opus file

nonemm-recorder is a separate program (src/Nonemm.Recorder.App) over a
library (src/Nonemm.Recorder). Each row in its list is one channel of one
device, and each row becomes one channel in an Ogg Opus file: 60 ms frames,
complexity 9, 32 kbps per channel by default.

- A device that is not connected keeps its row, in red, and writes silence.
  It is recorded from the moment it returns, also during a recording.
- Output devices are recorded as loopback: WASAPI on Windows, the
  "Monitor of" sources on PulseAudio and PipeWire.
- A headphone button plays one channel, in mono, on the default device.
- The rows, bitrate and folder are saved when the window closes.

SoundFlow 1.4.2 numbers its backend enum one higher than the miniaudio it
ships, so asking for PulseAudio gave ALSA and Wasapi gave DirectSound.
SoundFlowBackend lowers the value by one and checks which backend started.

Output devices and live capture are not tested on real hardware. The
tests use a fake sound system and decode the written file.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

README.md

@@ -40,6 +40,11 @@ Everything that judges a station — the bar under the callsign box, a row in th
 check window, a spot on the bandmap, a line in the log — is coloured by the same
 scorer: red for a dupe, green for a new multiplier, blue for points.
 
+## Sound recorder
+
+`nonemm-recorder` is a small separate program that records channels of several
+audio devices into one Opus file. See [docs/recorder.md](docs/recorder.md).
+
 ## Windows
 
 Build the Windows copy on any machine that has the .NET 10 SDK: