async dandiapi

Tooling for backup in https://github.com/dandi/dandisets/, in particular https://github.com/dandi/dandisets/blob/draft/tools/backups2datalad/adandi.py provides implementation of DANDI API helpers for async Python. I think, eventually (upon the first call for their reuse in some other projects) those should be moved from downstream project into some central library. Even now it could be beneficial since we do have already established integration testing of dandi-archive against dandi-cli, but not bother testing our backup scripts. That results in needing to know to adjust them (e.g. like https://github.com/dandi/dandisets/pull/277) whenever API behavior needs change in clients. So I think the time is right.

I have not researched on what is common approach - to provide a separate project (e.g. `adandi` on PyPI) and possibly separate codebase or have the same codebase providing both sync and async interfaces (and dependencies via `extra_requires`) .

@jwodder -- what do you think would be the best course of action?