Extra thanks to Anders, for pointing out that the likelihood ratio is the *ideal* test statistic, according to the Neyman-Pearson lemma. Also, extra thanks to Teruki, for sending a paper which shows exactly how the debinning function can be useful for fitting. It is too bad that I did not know enough about stats to understand the paper back in 2012 when I first read it.

Changes to be committed:
modified: examples/debinningDemo.py
modified: examples/discriminationMC.py