BUG/REF: ArrowExtensionArray non-nanosecond units

[lukemanley]

• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/v2.1.0.rst file if fixing a bug or adding a new feature.

Fixes a few bugs related to non-nanosecond units for pyarrow duration and timestamp types.

The underlying issue is upstream (apache/arrow#33321) where pyarrow does not handle pandas Timedelta and Timestamp non-nano units correctly.

Example (1 seconds -> 0 seconds):

In [1]: import pandas as pd

In [2]: import pyarrow as pa

In [3]: td = pd.Timedelta(1, unit="s").as_unit("s")

In [4]: pa.scalar(td, type=pa.duration("s"))
Out[4]: <pyarrow.DurationScalar: datetime.timedelta(0)>

In [5]: pa.array([td], type=pa.duration("s"))
Out[5]: 
<pyarrow.lib.DurationArray object at 0x10751d120>
[
  0
]

[mroeschke]

Is there a circular import if this is put at the top?

[lukemanley]

updated, this is now at the top

[jbrockmendel]

Is there some amount of this we can hope to rip out once pyarrow is updated?

[lukemanley]

Is there some amount of this we can hope to rip out once pyarrow is updated?

I think it can all be removed eventually. It may take some time as it needs to be fixed upstream first and then the pyarrow min version needs to catch up.

[mroeschke]

Looks okay. Just a merge conflict

[mroeschke]

Thanks @lukemanley