Make Cython in numpydev config Consistent

[WillAyd]

Follow up to #34711 and just matching the other config files. Not sure generally if we should be specifying here and in higher level configs as may be ambiguous as to what Cython version gets installed, but going with the quick patch for now

[WillAyd]

I guess another alternate is to hard code only this to Cython 0.29.19 and leave the higher level config alone (essentially reverting #34711) @TomAugspurger

[TomAugspurger]

Does the CI failure make sense to you?

____________________ TestDtype.test_check_dtype[Int64Dtype] ____________________

[gw1] linux -- Python 3.9.0 /home/travis/virtualenv/python3.9-dev/bin/python

self = <pandas.tests.extension.test_integer.TestDtype object at 0x7f7438777400>

data = <IntegerArray>

[   1,    2,    3,    4,    5,    6,    7,    8, <NA>,   10,   11,   12,   13,

   14,   15,   16,   17,...5,   86,   87,   88,   89,   90,   91,

   92,   93,   94,   95,   96,   97, <NA>,   99,  100]

Length: 100, dtype: Int64

    def test_check_dtype(self, data):

        dtype = data.dtype

    

        # check equivalency for using .dtypes

        df = pd.DataFrame(

            {"A": pd.Series(data, dtype=dtype), "B": data, "C": "foo", "D": 1}

        )

    

        # np.dtype('int64') == 'Int64' == 'int64'

        # so can't distinguish

        if dtype.name == "Int64":

            expected = pd.Series([True, True, False, True], index=list("ABCD"))

        else:

            expected = pd.Series([True, True, False, False], index=list("ABCD"))

    

        # FIXME: This should probably be *fixed* not ignored.

        # See libops.scalar_compare

        with warnings.catch_warnings():

            warnings.simplefilter("ignore", DeprecationWarning)

            result = df.dtypes == str(dtype)

    

>       self.assert_series_equal(result, expected)

pandas/tests/extension/base/dtype.py:84: 

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

pandas/tests/extension/base/base.py:13: in assert_series_equal

    return tm.assert_series_equal(left, right, *args, **kwargs)

pandas/_libs/testing.pyx:68: in pandas._libs.testing.assert_almost_equal

    cpdef assert_almost_equal(a, b,

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

>   raise_assert_detail(obj, msg, lobj, robj, index_values=index_values)

E   AssertionError: Series are different

E   

E   Series values are different (25.0 %)

E   [index]: [A, B, C, D]

E   [left]:  [True, True, False, False]

E   [right]: [True, True, False, True]

I think my preference is to just pin in the environment files and work to fixing the needs for the pins before 1.1 is released.

[WillAyd]

Surprising to see that at first glance. Just restarted let's see

[WillAyd]

Looks like that failure is showing up on a few other PRs

[WillAyd]

#34718 might be better than this for now. Let's see how that looks first