BUG: GH3311 Dataframe where with a datetimelike was not correctly selecting

[jreback]

closes #3311

[jbrockmendel]

@jreback I think this PR introduced some behavior (definitely codified it in a test) that is incorrect. From the test:

        df = DataFrame(dict(A = date_range('20130102',periods=5), 
                            B = date_range('20130104',periods=5),
                            C = np.random.randn(5)))

        stamp = datetime(2013,1,3)
        result = df[df > stamp]

But if you just look at the df > stamp:

>>> df > stamp
       A     B     C
0  False  True  True
1  False  True  True
2   True  True  True
3   True  True  True
4   True  True  True

Consistency with Series, Index, and scalars would require comparison with the "C" column to raise TypeError. Also the reverse inequality also comes back as all-True for the "C" column:

>>> df < stamp
       A      B     C
0   True  False  True
1  False  False  True
2  False  False  True
3  False  False  True
4  False  False  True

So the question: OK to make this raise? Or is there a compelling reason for this to be different?