How do we include python documentation in doc-strings?

Once you import a python model using Sckitlearn.jl the documentation is available:

julia> @sk_import linear_model: Lasso
PyObject <class 'sklearn.linear_model._coordinate_descent.Lasso'>

help?> Lasso
search: Lasso LassoRegressor LassoCVRegressor LassoLarsRegressor LassoLarsICRegressor LassoLarsCVRegressor

Linear Model trained with L1 prior as regularizer (aka the Lasso)

    The optimization objective for Lasso is::

        (1 / (2 * n_samples)) * ||y - Xw||^2_2 + alpha * ||w||_1

...

However, we do not want using MLJScikitLearnInterface to generate every python model to get this documentation, for inclusion in MLJ model docstrings. (Currently, the python objects are only generated at run time, when someone calls fit on the MLJ model; see here). So I guess we generate this documentation separately and have __init__ load it in? Then there is the problem of synchronisation. Can julia artefacts help here?

Start with src/macros.jl around line 141 to understand when Python objects are generated at runtime. Investigate how init and Julia artefacts could load separately generated documentation, and determine how synchronization would work. Done means MLJ model docstrings include the Python documentation without importing every model during using.