An initialized optimizer is a tensorflow object, which (at least in
graph mode in tf1) is not deepcopy-able. Even if we were able to
deeocopy it, we probably wouldn't want to since it contains state.
Scikit-learn needs to be able to deepcopy an estimators arguments so
that it can create copies and derivatives of it.

Instead we require the uninitialized optimizer and its parameters to be
passed to our learners separately. The learner can then initialize the
optimizer as needed.

Newer keras versions delegate to tf.keras and therefore need tf2. See
https://github.com/keras-team/keras/releases/tag/2.4.0.