ERR: need a better error message when trying to compare incompat types on merge

[jreback]

xref #11410 (comment)

These are both datetimes, but not compatible.

In [8]: df = DataFrame({'A':pd.date_range('20130101',periods=3),'B':pd.date_range('20130101',periods=3,tz='US/Eastern')})

In [9]: pd.merge(df, df, left_on='A', right_on='B')
TypeError: Argument 'values' has incorrect type (expected numpy.ndarray, got Index)