2d Order statistics is almost making total sense to me. I think I need to change the integration operators A_H and A_V to be much more simple. Also, it's clear when thinking about the probabilities that each nfrVrH will not be normalized to one. For instance, given a "crescent moon" distribution, the 10f10V5H function will be normalized to a very small number, since the overall probability of rV=10, rH=5 is itself tiny. I was able to verify this idea with the twoDPlayaround script.

Changes to be committed:
modified: debinningCore/histPlotting.py
modified: debinningMath/sampleFunctions.py
modified: examples/nfrVrHDemo.py
new file: examples/twoDPlayaround.py
new file: ipython_notebook_slides/orderStatistics.ipynb