Enable Opus ML improvements

[ann0see]

What is the current behaviour and why should it be changed?

We have updated opus to a version which supports some new - usually disabled improvements based on machine learning. https://opus-codec.org/demo/opus-1.5/

It's probably worth testing them - they should improve quality in case of packet loss.

@38github hinted at those in https://github.com/orgs/jamulussoftware/discussions/3244

I have personally quickly tried enabling them some time ago, but didn't succeed yet as some new headers need to be added (didn't have time debugging yet).

[stefan1000]

fyi xiph/opus#425 (afaik the ML improvements have been trained on voice and not hq music)

[ann0see]

Thanks for the heads up. We could still try enabling those in future though.