Return NotImpelmented for arith ops with object-dtype Index subclasses

[TomAugspurger]

This change in 1.0 impacted xarray

 In [22]: import xarray as xr

In [23]: import pandas as pd

In [24]: idx = pd.timedelta_range("1D", periods=5, freq="D")

In [25]: a = xr.cftime_range("2000", periods=5)

In [26]: idx + a
/Users/taugspurger/sandbox/pandas/pandas/core/arrays/datetimelike.py:1204: PerformanceWarning: Adding/subtracting array of DateOffsets to TimedeltaArray not vectorized
  PerformanceWarning,
Out[26]:
Index([2000-01-02 00:00:00, 2000-01-04 00:00:00, 2000-01-06 00:00:00,
       2000-01-08 00:00:00, 2000-01-10 00:00:00],
      dtype='object')

In [27]: a + idx
Out[27]:
CFTimeIndex([2000-01-02 00:00:00, 2000-01-04 00:00:00, 2000-01-06 00:00:00,
             2000-01-08 00:00:00, 2000-01-10 00:00:00],
            dtype='object')

cc @jbrockmendel @dcherian. Brock suggested checking type(other), and returning NotImplemented for index subclasses we don't know about, to give the subclass a chance to take over the op.

[jbrockmendel]

Should we try to get a short-term fix in for 1.0?

[TomAugspurger]

Yep, I think so.