Clarify kwargs in feature functions

Some of our feature functions (melspectrogram, mfcc, etc) use kwargs to pass through parameters to core functions (stft, filter construction etc). This isn’t terribly well documented, and can be a little difficult for new users to navigate and understand what the options are.

At this point, it’s probably worth replacing some of these kwargs, eg https://github.com/librosa/librosa/blob/25538adb3aed3485a06e60b6dad88be3d540f0c2/librosa/feature/spectral.py#L1767-L1769 by

def mfcc(
    *, y=None, sr=22050, S=None, n_mfcc=20, dct_type=2, norm="ortho", lifter=0,
    n_mels=128, n_fft=2048, ...
):

Alternatively, if we don’t want to make the API totally redundant, we could at least provide more documentation in the kwarg’ed functions’ docstrings to better explain what options are available.

[Embot2000]

If no one has started working on this issue I’d like to give it a try!

[bmcfee]

Thanks @Embot2000 - that looks like a good start. I might hold off on the last two since they’re kwargs to external functions in numpy and scipy.

On the feature side, I think you have it covered. We could also look at some other parts of the package after that, eg effects and decompose, but let’s get the feature parts done first.