[BUG] == or != between incomparable types raises TypeError instead of returning an object with True/False

[mroeschke]

Describe the bug

In [1]: import pandas as pd, cudf

In [2]: cudf.Series(["s", "b"]) == cudf.Series([1, 2])
...
    113 elif isinstance(data, abc.MutableMapping):
    114     # This code path is performance-critical for copies and should be
    115     # modified with care.
    116     if data and verify:
    117         # Faster than next(iter(data.values()))
--> 118         column_length = len(data[next(iter(data))])
    119         # TODO: we should validate the keys of `data`
    120         for col in data.values():

TypeError: object of type 'bool' has no len()

In [3]: pd.Series(["s", "b"]) == pd.Series([1, 2])
Out[3]: 
0    False
1    False
dtype: bool

Expected behavior
Same as Out[3]

Additional context
Column._binarop does return NotImplemented when encountering an incomparable type to dispatch to the reflected's implementation, but if the other side return NotImplemented then we get a scalar boolean which we are not expecting.

Generally if we can narrow the inputs to _binarop to a Column (scalars get broadcast to len of the column), then we should be more aggressively raise TypeError instead of return NotImplemented for invalid type comparisons.