Fix numpy warning

[TomAugspurger]

No description provided.

[jbrockmendel]

i dont think we want to float("nan")s in here. maybe this should be np.float64("nan")? best guess is the point of having this in addition to np.nan is bc nan is not a singleton like pd.NaT.

[TomAugspurger]

This is equivalent to what we had before, since np.float was an alias for float. You're saying that isn't what we want?

[jbrockmendel]

just speculating; im pretty confident we dont want two identical objects in here

[TomAugspurger]

Ah I missed that we already had float('"nan").

[TomAugspurger]

Various failures like

=================================== FAILURES ===================================

_______________ TestIndex.test_isin_nan_common_object[nan0-nan2] _______________

[gw0] linux -- Python 3.8.3 /home/travis/miniconda3/envs/pandas-dev/bin/python

self = <pandas.tests.indexes.test_base.TestIndex object at 0x7ff81c8b2e80>

nulls_fixture = nan, nulls_fixture2 = nan

    def test_isin_nan_common_object(self, nulls_fixture, nulls_fixture2):

        # Test cartesian product of null fixtures and ensure that we don't

        # mangle the various types (save a corner case with PyPy)

    

        # all nans are the same

        if (

            isinstance(nulls_fixture, float)

            and isinstance(nulls_fixture2, float)

            and math.isnan(nulls_fixture)

            and math.isnan(nulls_fixture2)

        ):

>           tm.assert_numpy_array_equal(

                Index(["a", nulls_fixture]).isin([nulls_fixture2]),

                np.array([False, True]),

            )

which I can't look into at the moment. I'll just remove the duplicate key.