[ENH] Dispatch aggregate, refactor AggJoiner & AggTarget

[TheooJ]

The goal of this PR is to dispatch aggregate, currently written in two files, by directly implementing it in _agg_joiner.py.

Following discussions with @Vincent-Maladiere and @jeromedockes, AggJoiner and AggTarget now require the operations parameter by default, and will try to apply all operations on all columns — as opposed to now, where columns are separated in categorical and numeric and only some operations are computed on each category.

I’m planning on doing follow ups to completely remove the _pandas.py, _polars.py, _namespace.py files, and on refactoring AggTarget with cross-fitting.

[Vincent-Maladiere]

Hey @TheooJ, very nice effort! Here is a first pass of comments.

On a higher level:

1. Since you use the CheckInputDataFrame class in AggJoiner.fit_transform, you should be able to remove the _check_dataframes method entirely. And since _check_inputs currently calls _check_dataframes, you have to place them in reverse order:

self._main_check_input = CheckInputDataFrame()
X = self._main_check_input.fit_transform(X)
self._check_inputs(X)

Additionally, the check_inputs method of the AggTarget could be simplified because CheckInputDataFrame does most of the checks.

2. You should add a get_feature_names_out method that returns self.all_outputs_ for both AggJoiner and AggTarget.

3. I think we should allow key, aux_key and main_key to be selectors.

[TheooJ]

If think the PR is good for a second pass. I probably need to rename it and add some extra things in the changelog.

@Vincent-Maladiere I addressed points 1 & 2: simplifying AggTarget's _check_dataframes (it was more of a complete RFC of AggTarget) and adding a get_feature_names_out method. Point 3 I didn't implement because I'm not sure it's a good idea to indicate users "you should use selectors to choose keys", so I didn't implement it in this PR. Shouldn't be a blocker for the review though and it's a quite minor change IMO

[Vincent-Maladiere]

Nice effort! I think we are almost done here

[Vincent-Maladiere]

Should we clarify what selector means?

Suggested change

	cols : str or iterable of str or selector, default=s.all()
	cols : str or iterable of str or skrub selector, default=s.all()

[jeromedockes]

at this point the _selectors module is still private right? maybe it should be made public if we document this here?

[TheooJ]

Wonder how @GaelVaroquaux feels about this
For context, we allowed cols to be a selector in AggJoiner and AggTarget

[Vincent-Maladiere]

It might be a good time to start writing the dev doc in another PR and fixing #991

[Vincent-Maladiere]

Sorry, there's some duplicate with Jerome's feedback, we reviewed it simultaneously

[Vincent-Maladiere]

Notwithstanding the discussion regarding the documentation of skrub selector (which may be done in another PR), this LGTM! Congratulations @TheooJ :)

Let's aim to remove _namespace.py, _polars.py, and _pandas in a follow-up PR!

[TheooJ]

After discussion with @GaelVaroquaux, it was decided we won't make selectors public here. Let's open a meta issue on their support and documentation first

Let's aim to remove _namespace.py, _polars.py, and _pandas in a follow-up PR!

I'm on it :)