BUG: Timestamp comparisons to dataframe should raise if there are mismatched types

[shoyer]

Compare (on Python 2.7):

In [7]: pd.Timestamp('2000-01-01') > pd.Series(range(5))
TypeError: Cannot compare type 'Timestamp' with type 'int'

In [8]: pd.Timestamp('2000-01-01') > pd.DataFrame({'x': range(5)})
Out[8]:
      x
0  True
1  True
2  True
3  True
4  True

Probably will fix this in the process of getting #8916 to pass, but I wanted to raise this as a separate issue in case this is intentional. Pretty sure it's not, but there was a test-case specifically introduced that relies on this behavior (see PR #4983).

[jreback]

this is related to this: #8554

Its actually a bit non-trivial as the scalar/Index/Series machinery is well tested/consistent. However, the DataFrame machinery for the same takes a different path and thus needs to be changed (This is the whole usage of Timeops in core/ops.py.