Verbose debugging for spectral clustering

[sstalley]

Describe the workflow you want to enable

I want to be able to enable verbose debugging messages for spectral clustering to make it easier to find problems and debug.

Describe your proposed solution

Add a 'verbose' optional argument to the spectral_clustering() function that gets passed into the underlying k_means() call.

Describe alternatives you've considered, if relevant

There may be additional places in the spectral clustering algorithm where it would make sense to add a debug message.
That said, just getting the messages from k means was enough for me to debug my problems, and is a big step up from no print messages at all.

Additional context

I have been using sklearn for the past year or so and would like to learn how to contribute to the project.
I noticed that spectral clustering was missing verbose debugging (a feature present in a lot of the other clustering algorithms in sklearn), and I thought that adding this feature would be a good way to get familiar with the sklearn workflow.

Mostly I am testing the waters to see if this feature is something that would be valuable to the community and accepted by the maintainers.
I am a former software engineer (and current ML research assistant) who has contributed to open source projects in the past (like wireshark and the linux kernel).

[thomasjpfan]

We are in the process of rethinking the we do logging: #17439

That being said, passing in verbose through to kmeans does make sense. Although we most likely need one logging messaging in spectral_clustering stating that it is assigning labels with kmeans.