Support `async` for bulk operations

Lots of what we do in `_to_apply` here is either going to be disk (in the case of `Path`) or network (if we implement #1) bound and should benefit from doing it async.

https://github.com/drivendataorg/pandas-path/blob/8b7d1478966084733df09864b680b1eec01da0d8/pandas_path/accessor.py#L77-L105

Start by reading pandas_path/accessor.py at the linked _to_apply section (lines 77–105) and trace how bulk operations are currently executed. The issue does not define the async API, supported operations, or tests; clarify those requirements before implementation, with done meaning that the agreed bulk operations support async execution.

Written by the indexing model from the issue text.