Added very basic silhouette score calculation to CAP

[m-miedema]

As an example relevant to #43, this is a very simple implementation that would give access to silhouette scores. I think this is more informative than simply returning the mean silhouette score, although it would probably be helpful to give a simple plotting example in a demo notebook.

[m-miedema]

I'm a bit tangled in the weeds of previous commits upstream here - let's discuss and then I can rebase this PR appropriately :)

[mtorabi59]

Hi @m-miedema , thank you so much for implementing this! I think overall this can be useful for PydFC.
Maybe we could discuss about the way it is implemented. For something similar to the current version, maybe it could've been easier if the user was calculating the silhouette score outside CAP and without having it inside CAP implementation, for instance:

measure = CAP(**params_methods)
measure.estimate_FCS(time_series=BOLD_multi)
silhouette_score = silhouette_samples(time_series.data.T.astype(np.float32), measure.Z)

Unless we want to implement a function or method of CAP class that searches for best number of states on its own.
What do you think?