ENH: support CategoricalIndex Followups

[jreback]

followup to #9741

These are all commented in pandas/core/test_frame/test_categorical_indexing_index

• support slice semantics
• fix index name issue on selection for non-existant label that IS in the categories
• fix index name issue on assignment for non-existan label that IS in the categories
• duplicated returns Boolean array? / test
• in repr, use max_categories/max_seq_len to display a truncated view

These last 2 were 'solved' by putting the following in core/internals/reindex_indexer
but this caused some issues as the MultiIndex._shallow_copy is really a view semantic (so that needs to be fixed as well)

# make sure that we propogate the correct meta data for this axis
# we need to do this because we are not doing a take / reindex
# which would do this automatically
new_axis = self.axes[axis]._shallow_copy(new_axis)

In [1]: In [4]: df = DataFrame({'A' : np.arange(6,dtype='int64'),
   ...:    ...:                 'B' : Series(list('aabbca')).astype('category',categories=list('cabe')) }).set_index('B')

In [2]: df
Out[2]: 
   A
B   
a  0
a  1
b  2
b  3
c  4
a  5

In [3]: df.loc['e'] = 20

In [4]: df
Out[4]: 
    A
a   0
a   1
b   2
b   3
c   4
a   5
e  20

[TomAugspurger]

This is tangentially related. I think that the .categories object should return a CategoricalIndex instead of a regular Index like is does now. My need right now is to be able to reindex a categorical with a .categories. We decided to have .reindex(Index) have an Index and .reindex(CategoricalIndex) have a CategoricalIndex. Should I open a separate issue? That should be a quick fix.

[jreback]

@TomAugspurger that sounds right. I thought we discussed that in another issue (but can't find right now).

[TomAugspurger]

Ok, I'll do that now since it will simplify my categorical value_counts code a lot.