feat: Add --write-encoder-settings to tag streams with encoder settings

[ppetr]

When enabled, the second-pass command writes an ENCODER_SETTINGS tag on each normalized audio stream, holding an ffmpeg-command-equivalent string with the codec, bitrate, sample rate and channels plus the normalization filter including the measured loudnorm values. It is added to the existing second-pass ffmpeg invocation, so no extra pass is performed.

Has no effect together with --metadata-disable. CLI docs, shell completions and tests updated.

[slhck]

Thanks, this looks reasonable and is fairly complete. I'd just like to hear your motivation or use case for it?

[ppetr]

Hi! I have this little tool https://github.com/ppetr/audio-to-opus to convert audio to low bandwidth Opus files. While experimenting with various flag to optimize the codec for speech, I realized it'd be quite convenient to be able record the respective flags so that I could see what recording was done with what flags.

And then I thought that a better place for this improvement would be ffmpeg_normalize so that others could benefit from it too (and also it has all the necessary information already available for the 2nd pass, so no 3rd remux is needed).

[slhck]

This is now published, thanks!