DEPS: drop np 18 supoort

[fangchenli]

No description provided.

[fangchenli]

=================================== FAILURES ===================================
______________________________ test_where_dt64_2d ______________________________
[gw1] linux -- Python 3.8.0 /usr/share/miniconda/envs/pandas-dev/bin/python
    def test_where_dt64_2d():
        dti = date_range("2016-01-01", periods=6)
        dta = dti._data.reshape(3, 2)
        other = dta - dta[0, 0]
        df = DataFrame(dta, columns=["A", "B"])
        mask = np.asarray(df.isna())
        mask[:, 1] = True
        # setting all of one column, none of the other
        expected = DataFrame({"A": other[:, 0], "B": dta[:, 1]})
>       _check_where_equivalences(df, mask, other, expected)
pandas/tests/frame/indexing/test_where.py:1019: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
df =            A          B
0 2016-01-01 2016-01-02
1 2016-01-03 2016-01-04
2 2016-01-05 2016-01-06
mask = array([[False,  True],
       [False,  True],
       [False,  True]])
other = <TimedeltaArray>
[
['0 days', '1 days'],
['2 days', '3 days'],
['4 days', '5 days']
]
Shape: (3, 2), dtype: timedelta[64](https://github.com/pandas-dev/pandas/runs/6739385945?check_suite_focus=true#step:9:65)[ns]
expected =        A          B
0 0 days 2016-01-02
1 2 days 2016-01-04
2 4 days 2016-01-06
    def _check_where_equivalences(df, mask, other, expected):
        # similar to tests.series.indexing.test_setitem.SetitemCastingEquivalences
        #  but with DataFrame in mind and less fleshed-out
        res = df.where(mask, other)
>       tm.assert_frame_equal(res, expected)
E       AssertionError: Attributes of DataFrame.iloc[:, 0] (column name="A") are different
E       
E       Attribute "dtype" are different
E       [left]:  int64
E       [right]: timedelta64[ns]
pandas/tests/frame/indexing/test_where.py:990: AssertionError

[jreback]

thanks @fangchenli

I wouldn't mind actually bumping to 1.20+ per: https://numpy.org/neps/nep-0029-deprecation_policy.html

but no big deal to keep here as well.