Update cl_bayes.py with GridSearchCV and KNN

[CCLynch]

I've refactored the 'cl_bayes.py' file to address #20 and added GridSearchCV for some of the models.

• 'plot_clf_boundary()' should work with any estimators on a 2d dataset. So if you want to plot the decision boundary and a scatter of the true/false pos/neg points, it should be pretty plug-n-play.
• I also added KNN as a 4th model, so now it's LDA, QDA, GaussianNB, and KNN.
• With GridSearchCV, there's one forbidden combination of solver and shrinkage that will throw an error, but solving it would require writing the param grid from scratch, which I'm not going to do at the moment, since it doesn't affect the model outcome or score calculation.

I still need to write 'lda.md', and I will probably change the name to reflect that it's multiple bayesian models. I'll start that after this gets merged.

[sophiacofone]

Looks good, thanks for taking on #20 and adding knn, i would be curious to see how knn and these other models due with the euc. distances but that can always be for future work or a side thing.

I agree that the lda.md needs to be written/name changed. Even if you dont want to write it all up right now, it might be nice to add a make target to the makefile and point to it in that file so we can easily reproduce the figures. Otherwise looks good