Rolling back the effects experiment

The January 15 audio history is a good example of why a development journal needs the rollback as well as the new-feature commit. Equalizer, noise-suppression and automatic-gain-control work was added, then explicitly reverted later that day. Those features shouldn't appear in a current capabilities list just because their introduction is still in Git.

The next commit simplified the supported formats and topology. It removed the extra surround expansion and unused bass, treble, reverb, chorus and echo-cancellation nodes and related handlers. The direction was a smaller surface that would be easier to understand and validate.

There is a wrinkle worth explaining in plain terms. The simplification commit's description is narrower than all of the retained table entries. The current microphone table still contains a floating-point example, and the speaker table retains additional entries beyond its headline range comments. A clean-sounding commit summary isn't an authoritative compatibility matrix.

For this entry, I'm describing the removal and the remaining need to verify format negotiation rather than publishing a new exact list of supported combinations. That list needs to come from the actual advertised ranges, table entries and successful stream-open tests together.

The same care applies to the README, which still includes broader historical descriptions. Updating the documentation against the implementation is useful next work, alongside native Windows tests for both endpoints and the relevant signing path.

What I like about recording this stage honestly is that it makes the next decision easier. We know what was tried, what was reverted and where the remaining ambiguity is. We can test a focused driver instead of carrying forward a feature list assembled from several different generations of code.