Bayesian optimization tutorial

[hurricane642]

In this PR I add an example of using Bayesian optimization for Power Spectrum. Most of the code is based on the first librascal example, I have not changed the meaningful part of it, adding only a few things about the optimization itself. This is a fairly simple example with the simplest functions from the scikit-optimize library, but I hope it is enough for the first acquaintance. This is my first time writing examples for the library, so any suggestions and edits are welcome.

[max-veit]

Very good, thanks for adding this! I probably won't have time for a full review anytime soon (maybe late next week but no promises). But right now my main comment is that there's a lot of utility code, especially in the second-to-last cell, that would benefit from either being moved to its own "notebook utilities" module or broken down and explained step-by-step in the notebook.

BTW the tests are failing because of an update in scikit-cosmo a while back; there's a fix in 95aedf9

[hurricane642]

Very good, thanks for adding this! I probably won't have time for a full review anytime soon (maybe late next week but no promises). But right now my main comment is that there's a lot of utility code, especially in the second-to-last cell, that would benefit from either being moved to its own "notebook utilities" module or broken down and explained step-by-step in the notebook.

BTW the tests are failing because of an update in scikit-cosmo a while back; there's a fix in 95aedf9

Thank you so much for your reply! I've tried to account for the errors, added more comments on how the code works and broken it down into separate parts. I wanted to clarify what you meant about the test. That is, how do I make the tests pass?:)