Deal with grouped/ungrouped DataFrames in Steps

[prockenschaub]

Addresses the issues described in #13

The approach taken here is to use a function _check_ingredients that a) checks if a step accepts grouped dataframes and throws an error if it doesn't and b) removes the group if asked to via the remove_group parameter. See individual Steps for details.

[HendrikSchmidt]

Could you add typehints and docstrings for the added methods? (And maybe ideally for every touched method, that way we'll get there little by little for the 'legacy' code as well.)

[prockenschaub]

@HendrikSchmidt I've added docstrings and typehints, thanks for pointing this out.

[HendrikSchmidt]

Would it make sense to refactor new_data = self._check_ingredients(data, remove_group=True) into a similar transform and do_transform pattern like with fit?

[prockenschaub]

Note that my test for impute fill introduced a recipe with NaN but such a recipe already seems to exist in another branch in conftest.py. I need to resolve this.