Plot functions from `OptimizeResult` object about scikit-optimize

[glouppe]

I'll work on plot_convergence, this would be convenient for all our notebooks. I am thinking about designing it so that plot_convergence can either accept a single OptimizeResult object, a list of them (hence plotting the average and the individual traces), or a list of lists of them.

[betatim]

In addition to the GPyOpt plots it would be good to have something to show where the samples are for N>2 dimensions. Maybe a corner.py style plot.

[glouppe]

What could be another interesting plot? in GPyOpt they have this plot which shows the distance between consecutive points. This is nice to study the exploitation/exploration trade-off, no matter the dimension of the search space. The caveat is that not all distance metrics are appropriate...

[betatim]

Listing some ideas:

• #67 (comment) I'd love this or other tools to help me see if the optimiser got stuck/is doing the right thing.
• I also found the 2D plot from @kegl useful as well as the 1D-convex-hull one. plots towards bottom of the ipynb

[glouppe]

Closing this, we have some nice ones now :)