PERF: numexpr doesn't support floordiv, so don't try

[jorisvandenbossche]

See overview of supported operators: https://numexpr.readthedocs.io/projects/NumExpr3/en/latest/user_guide.html#supported-operators. By trying to use it, we were falling back to the (slower) _masked_arith_op

Discovered by investigating the arithmetic benchmarks at #39146 (comment)

For example using the arithmetic.IntFrameWithScalar.time_frame_op_with_scalar(<class 'numpy.float64'>, 3.0, <built-in function floordiv>) benchmark:

import operator

dtype = np.float64
arr = np.random.randn(20000, 100)
df = pd.DataFrame(arr.astype(dtype))
scalar = 3.0
op = operator.floordiv

In [2]: %timeit op(df, scalar)
56.4 ms ± 723 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)  <-- master
24.3 ms ± 236 µs per loop (mean ± std. dev. of 7 runs, 10 loops each)  <-- PR

[jbrockmendel]

LGTM (im assuming the CI failures are unrelated MPL stuff)

[jorisvandenbossche]

There is still a failure in test_expressions.py to fix

[lithomas1]

@jreback Looks like this was red when merged and should be reverted.

e.g.

=================================== FAILURES ===================================
________ TestExpressions.test_bool_ops_raise_on_arithmetic[//-floordiv] ________
[gw1] linux -- Python 3.7.1 /usr/share/miniconda/envs/pandas-dev/bin/python

self = <pandas.tests.test_expressions.TestExpressions object at 0x7fc5e381b630>
op_str = '//', opname = 'floordiv'

    @pytest.mark.parametrize(
        "op_str,opname", [("/", "truediv"), ("//", "floordiv"), ("**", "pow")]
    )
    def test_bool_ops_raise_on_arithmetic(self, op_str, opname):
        df = DataFrame({"a": np.random.rand(10) > 0.5, "b": np.random.rand(10) > 0.5})
    
        msg = f"operator {repr(op_str)} not implemented for bool dtypes"
        f = getattr(operator, opname)
        err_msg = re.escape(msg)
    
        with pytest.raises(NotImplementedError, match=err_msg):
>           f(df, df)
E           Failed: DID NOT RAISE <class 'NotImplementedError'>

from https://github.com/pandas-dev/pandas/runs/2245866400

[jreback]

whoops

hard to tell sometimes

ok

[jreback]

reverted #40742

[jorisvandenbossche]

There was even a comment that I still needed to fix the test .. ;)