Add RxC polarization functions

[gabeschoenbach]

Added the RxC polarization functions, and then provided tests for both the 2x2 and RxC versions.

RxC polarization function:

• Requires two more arguments than the 2x2: the groups the user is interested in, and the candidate upon which to compare preferences.

Changes to the 2x2 polarization functions include:

• Fixing the probability/percentile issue — np.percentile() takes in a number between 0 and 100, not 0 and 1, so changed the arguments to percentile instead of probability and adjusted the arithmetic inside the function to give the correct results.
• Generalized the group and group_complement to groups[0] and groups[1] to better match the behavior of the RxC function.
• Updated the plot_polarization_kde() function to accept groups so that it can be called on either the 2x2 or RxC polarization reports.

Tests:

• Tested the plot_polarization_kde() function just by checking a Matplotlib Axis is returned.
• Tested the output of polarization_report(): increasing the threshold should decrease the probability and decreasing the probability should increase the threshold.

Also updated the .pylintrc config file to not ignore comments, since I have some duplicate code that should pass the linting. So then I just added some small comments around this duplicate code to get it to pass — a bit of a hack. Finally, added examples in both the 2x2 and RxC Santa Clara demo notebooks!

[karink520]

Lots of good stuff here - I like the switch to groups[0], groups[1] and think there will be more we can do down the line to keep making RxC and 2x2 more compatible (we're probably due for a bigger refactor there at some point, but I want to wait until we have the last couple methods implemented first). Glad you caught and fixed the percentile/probability issue, and thanks for putting in a corresponding test too!