Still to do:
* Figure out why silence for any reasonable length of time results in sadness for the note finder.

Try this:
* Separate "LED Selection" from "output" algorithms.
* For light finding, pick lights off the peaks to get number of lights to use.
* For light shifting (for 1d-looping light systems) shift the centers of the notes, then vernoi between the notes.
* Consider running DFT on all channels and mixing results