Feature/add uplift example

[paulf81]

Feature or improvement description This pull request adds an additional example demonstrating the calculation of total uplift. I think it's mostly ready but there are a few finer points I'd like to work through before merging:

• [x] What are the warnings about? Can they be squashed
• [x] Should we lock in a random seed to guarantee one result so the documentation can be more specific?

Related issue, if one exists Closes Issue #138

[misi9170]

Nice! On your points above:

• I'm strongly in favor of adding np.random.seed(0) right after if __name__ == "__main__": for reproducibility
• Did you mean to leave plt.show() off? There are a lot of plots made in this script that are abandoned
• I'm still looking into the warnings, but they all originate from the plot_binned_mean_and_ci call, so if you don't want to plot anything anyway that line can be commented out and the warnings don't appear. On the other hand, it's probably best to track those down and see what causes them?

EDIT: I've now fixed the warnings, I think. In doing so, I made the following changes:

• Renamed examples 05 and 06, because there was a double-up at 05 (likely my fault when reorganizing the examples)
• Added random seeds to both 06 and 07 (new numbering)
• Specified the x_edges passed to plot_binned_mean_and_ci(), which avoids empty bins caused by default edges
• Added handling for bins with no spread to plot_binned_mean_and_ci(), which is possibly not needed for realistic data but avoids warnings raised with the synthetic data that is too uniform

This fixes all warnings that I was seeing.

[paulf81]

Thanks @misi9170 ! I know you're making some last checks but feel free to merge when done, thank you!