Integrate OrdinalEncoder Component into AutoML

[tamargrey]

Once #1389 is finished, the OrdinalEncoder will exist but not actually be used within EvalML. We should integrate it.

This will require a change to the OneHotEncoder, since when the ordinal encoder is used, the one-hot encoder should not encode ordinal columns.

[tamargrey]

a thought I didn't want to forget: we use sklearn's OrdinalEncoder in the time series featurizer - maybe it'd be worth using ours instead, but then we'd need to change the ordinal encoder to use non ordinal categorical columns, so that's probably not what we want to do