UX: add hints to YTUnidentifiedDataType

[neutrinoceros]

Bug report

Bug summary

YTUnidentifiedDataType is the error ones get from yt.load(...) when no "valid" frontend is found. Currently the error message doesn't give much guidance, but in many a case it can be fixed by either:

• installing missing requirements (most often h5py), because many implementations of Dataset._is_valid() just return False when getting an import error.
• upgrading yt (if the current version is older than the first one to have the required frontend)

It would be helpful to acknowledge these solutions in the message. Ideally, I think we should avoid importing stuff in _is_valid methods, but that calls for a broad refactor with a high chance of breaking someone's workflow, so I think it's a little late to do that.

I note that #4272 is helpful here because it gathers the important data (frontend VS requirements mapping) in one place (pyproject.toml).