MI calculations with adaptive binning

The attached code implements a greedy heuristic for finding good binnings for matrix-based MI estimation.

Empirical evaluation of the method can be found in

I. Nelken, G. Chechik, T.D. Mrsic Flogel A.J. King and J.W.H. Schnupp
Encoding stimulus information by spike numbers and mean response time in primary auditory cortex.
J. Computational Neuroscience 19(2):199-221, 2005