Increase microtonal polyphony

Hi, I described the idea at Micro Scale - Midihub Documentation - #3 by uth but then I found out the forum “Feature requests”. See that linked topic for intro.

I would add here some implementation details and further optimisations. Every of the 127 frequencies would scan for its “octave neighbours”, that is, whose frequency ratio precision would fall below some small epsilon value specified in the pipe parameter. Then, the ± n*12 tone MIDI mechanism might be used if applicable, instead of merely the channel-hungry pitch bend.

In this way, the algorithm would detect and take advantage of any micro temperament featuring octaves. Otherwise, it would just behave like now.

The above would work for temperaments like 15-tone equal or inequal, but there is one more optimisation possible from which would additionally profit any n*12-tone equal temperament, like the 24-note quarter-tone temperament. Namely, not only “octave neighbors” would be searched for, but any “ratio which exists in 12-tone” neighbours. Thanks to this, in the case of quarter tone, only two channels would be enough to use the full device polyphony. For the 36-note equal temperament, 3 channels would be enough.