is_scalar should return True for DateOffset objects

[jbrockmendel]

[databasedav]

I am attempting this as a first issue.

I traced the is_scalar function back to its implementation which is in Cython and it uses checks imported from the CPython datetime library to check the for the listed datetime objects. DateOffset objects are specific to Pandas so what would my next steps be here? I imagine importing DateOffset from where ever it's defined and using isinstance is not the answer...

[jreback]

need to add a is_offset to pandas/_libs/src/inference.pyx, then you can use this in isscalar. this would look very similar to pandas/_libs/src/util.pxd/is_period_object (you don't need to define it also in util.pxd though).

is_offset should also be imported / tested in pandas/core/dtypes/common.py