The discrimination script now shows that (at least for one example) the debinning chisq statistic always has more power to exclude pure bg than logLike.

Next, try some fitting, and try other example s+bg distributions.

Changes to be committed:
modified: debinningCore/__init__.py
modified: debinningCore/debinning.py
modified: debinningCore/vsHistogram.py
modified: examples/debinningDemo.py
modified: examples/discriminationMC.py
modified: examples/zeroVarianceLimit.py