Metalearners

------------------

* No longer raise an error if ``feature_set`` is provided to
:class:`metalearners.SLearner`.

* Fix a bug where base model dictionaries -- e.g. ``n_folds`` or
``feature-set`` -- were improperly initialized if the provided
dictionary's keys were a strict superset of the expected keys.

------------------

* Ship license file.

------------------

* Fix dependencies for pip.

------------------

* Implemented :meth:`metalearners.cross_fit_estimator.CrossFitEstimator.clone`.

* Added ``n_jobs_base_learners`` to :meth:`metalearners.metalearner.MetaLearner.fit`.

* Renamed :meth:`metalearners.explainer.Explainer.feature_importances`. Note this is
a breaking change.

* Renamed :meth:`metalearners.metalearner.MetaLearner.feature_importances`. Note this
is a breaking change.

* Renamed :meth:`metalearners.explainer.Explainer.shap_values`. Note this is
a breaking change.

* Renamed :meth:`metalearners.metalearner.MetaLearner.shap_values`. Note this
is a breaking change.

* Renamed :meth:`metalearners.metalearner.MetaLearner.explainer`. Note this is
a breaking change.

* Implemented ``synchronize_cross_fitting`` parameter for
:meth:`metalearners.metalearner.MetaLearner.fit`.

* Implemented ``cv`` parameter for :meth:`metalearners.cross_fit_estimator.fit`.

------------------

* Implemented :class:`metalearners.explainer.Explainer` with support for binary
classification and regression outcomes and discrete treatment
variants.

* Integration of :class:`metalearners.explainer.Explainer` with :class:`metalearners.metalearner.MetaLearner`
for feature importance and SHAP values calculations.

* Implemented model reusage through the ``fitted_nuisance_models`` and ``fitted_propensity_model``
parameters of :class:`metalearners.metalearner.MetaLearner`.

* Allow for ``fit_params`` in :meth:`metalearners.metalearner.MetaLearner.fit`.

------------------

Beta release with

* :class:`metalearners.DRLearner` with support for binary
classification and regression outcomes and discrete treatment
variants.

* Generalization of :class:`metalearners.TLearner`,
:class:`metalearners.XLearner` and :class:`metalearners.RLearner`
to allow for more than two discrete treatment variants.

* Unification of shapes returned by ``predict`` methods.

* :func:`metalearners.utils.simplify_output` and :func:`metalearners.utils.metalearner_factory`.