Increased evolution speed (and slowed calculations between evolutions a little)
by adding a simple measure of an audivolv.Func to filter out
the obviously bad evolved Java codes that are unlikely to play sound.
It is FuncHome.howMuchMovementInEachDimensionOfHypercubeVectorField(Func).
When you click the 'Sounds Good' or 'Sounds Bad' button,
it keeps evolving new Java codes until 1 has movement scores
at least around the average movement scores (including Java codes not used).
Movement scores are individual to each index in the audivolv.Func's floating point array.
If too many movement scores are too low, compared to decaying averages,
the whole audivolv.Func fails (but you still hear it while its evolving),
and it is scored low, and things it was created from get scored low recursively.
It learns to fail that way less often.

Improved audivolv-path parsing to use quotes (not tested well),
like /classpath/anyfiles/"version history summary"/Audivolv0.1.4.txt
so files display in the window instead of an error.