unify freq strings (independent of pd version)

[mathause]

• Adresses points 2 and 3 and closesmore frequency string updates? #8612
• Tests added
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

Probably not ready for review yet.

[mathause]

Some more updates. I think this needs more work still. But to make sure I don't forget: I don't update the following line:

xarray/xarray/core/groupby.py

Line 330 in 5a92d48

loffset=pd.tseries.frequencies.to_offset(loffset)

There is a test with a negative frequency ("-12H") and to_offset cannot handle this (the regex does not allow for it.)

Should to_offset be able to handle negative offsets?

[spencerkclark]

Thanks @mathause — it does seem reasonable to update our to_offset regex to allow for the minus sign, if it is not too much trouble (the pandas version allows it).

[mathause]

I am marking this ready to review.

[spencerkclark]

Thanks @mathause—I think this looks good. Awesome that this enables us to silence all the frequency-string-related deprecation warnings in the test suite too!

I'm approving, but could you just add an additional test for both _legacy_to_new_freq and _new_to_legacy_freq that a pandas-only offset string gets passed through unmodified without raising an error?

[max-sixty]

Thanks a lot @mathause !