PyChrm: Additional FeatureSet_Discrete.Split() functionality

[GoogleCodeExporter]

1. Provide ability to (as also exists in C-Chrm) to leave a class out of 
consideration if it does not meet i & j class size criteria.

2. Provide ability to match up trainingset from one channel to that of another 
channel using string closest match provided by the core package difflib. 
Example: deconvolved H & E channel images with _H.tif & _E.tif suffixes. If I 
create a random train/test split for one channel, make so that you can create 
the same train test split for the other channel.

Original issue reported on code.google.com by [email protected] on 18 Jan 2013 at 9:11