[ENH] Add Probabilistic boosting and stacking compositors

[arnavk23]

Reference Issues/PRs

Towards #7

What does this implement/fix? Explain your changes.

• Adds ProbabilisticStackingRegressor and ProbabilisticBoostingRegressor as composable pipeline elements for probabilistic ensembling.
• Both compositors support mixture-based output, true residual-based boosting, and are extensible for meta-learners and advanced weighting.

Does your contribution introduce a new dependency? If yes, which one?

No new dependencies are introduced.

What should a reviewer concentrate their feedback on?

• API design and extensibility for future meta-learners and weighting schemes.

Did you add any tests for the change?

Yes.

Any other comments?

PR checklist

For all contributions

• I've added myself to the list of contributors with any new badges I've earned :-)
  How to: add yourself to the all-contributors file in the skpro root directory (not the CONTRIBUTORS.md). Common badges: code - fixing a bug, or adding code logic. doc - writing or improving documentation or docstrings. bug - reporting or diagnosing a bug (get this plus code if you also fixed the bug in the PR).maintenance - CI, test framework, release.
  See here for full badge reference
• The PR title starts with either [ENH], [MNT], [DOC], or [BUG]. [BUG] - bugfix, [MNT] - CI, test framework, [ENH] - adding or improving code, [DOC] - writing or improving documentation or docstrings.

For new estimators

• I've added the estimator to the API reference - in docs/source/api_reference/taskname.rst, follow the pattern.
• I've added one or more illustrative usage examples to the docstring, in a pydocstyle compliant Examples section.
• If the estimator relies on a soft dependency, I've set the python_dependencies tag and ensured
  dependency isolation, see the estimator dependencies guide.

[JiwaniZakir]

In _predict_proba of ProbabilisticStackingRegressor, when meta_learner_ is present, the code constructs Mixture(distributions=[("meta", meta_pred)], weights=[1.0]) where meta_pred is either a raw numpy array from predict_proba or a 1D array from predict — neither is a skpro distribution object. Mixture expects its distributions argument to contain actual distribution instances, so this path will almost raise a runtime error and appears untested by the get_test_params scenarios (which don't exercise the meta_learner path at all).

Additionally, _fit builds meta-features via est_fitted.predict(X).values.flatten() (point predictions), which discards all distributional information when training the meta-learner. For a "probabilistic stacking" regressor, you'd typically want to include at minimum both the predicted mean and variance (or quantiles) as features, otherwise the meta-learner has no basis for producing calibrated uncertainty estimates.

The add_base_estimator method mutates self.estimators in-place post-construction, which conflicts with skpro/sklearn's convention that constructor parameters are not mutated — cloning the estimator after calling this method would not preserve the added estimator, breaking pipeline serialization and cross-validation workflows.

[arnavk23]

Saw the same after my recent commit, trying to fix it. But thanks for the in detail review.