Add support for clustering

[rxavier]

Things to look out for

• y type hints should be modified
• check_estimator_type should be modified
• TargetEncoder depends on y so it shouldn't be used
• Maybe include some dimensionality reduction?

[rxavier]

It's not clear this is reasonable within Poniard's logic, as CV is not super straightforward to think about with clustering, and some sklearn clusterers do not have predict (OPTICS for example). See scikit-learn/scikit-learn#4301 scikit-learn/scikit-learn#20773 scikit-learn/scikit-learn#6154

[rxavier]

What would make sense is to have an interface to run different clusterers and score them on various metrics without CV. This would require overriding at least fit since we won't use cross_validate, predict and friends, setup would need to be fixed to allow y=None, show_results, etc.

Alternatively, the current BasePoniardEstimator could be used only for supervised problems, and a new more general base estimator made available

[BradKML]

If you guys are adding clustering how about also classification and PCA-like operations? Those might be useful as well

[rxavier]

Hey @BrandonKMLee. Classification is a core use case that's been implemented from the start, or do you mean something else?

As for PCA, you're free to use your own transformation pipeline instead of the default one (docs).