out of the box comparison with aupimo's benchmark

[jpcbertoldo]

I'll be PR-ing a tutorial notebook where i fetch aupimo scores from AUPIMO's benchmark (available in my github repo)

You can check it here, cell "In [25]".

As a user tests a new model on one of the those benchmark datasets, it would be useful to have a function that fetches those results and compares the user's scores to them out of the box.

This could also be integrated in the engine I guess.