Polynomial compressor

[cjrolo]

This PR lands the final compressor(s) for the BRRO compressor.

• Polynomial with two variants
  • IDW (Inverse Distance Weights) that should be better for more spiky signals
  • Linear/Catmull–Rom spline that should produce a continuous, easy to calculate curve over signals (vs a normal n-term polynomial)

Included are some changes to error handling (There was a bug in the error function) and minor fixes on FFT and conversion functions.

[rukai]

I've just noted a few things that can be simplified, go ahead and merge when ready.

[cjrolo]

Great review, thanks! I knew this would be a code that would need a good number of changes, and thanks for taking the time to review it.

@joshuabvarghese your tests saved me a great time here, detected a lot of things that could had go south very fast!