Improve error handling in SymmetryData

[mkhorton]

Hi everyone, just a small PR:

The SymmetryData model currently has optional fields, and therefore can be constructed even if symmetry information is unknown, e.g. if spglib fails.

This PR improves error handling in .from_structure to gracefully handle cases where spglib cannot determine the symmetry of a structure. A common use case where this might be encountered is if an otherwise valid StructureMeta (or a subclass of StructureMeta) is instantiated, but with symmetry information unknown.

The PR also improves handling of angle_tolerance to ensure that this is picked up from EmmetSettings instead of assuming the default value, and records this alongside symprec in the document model. This is also an optional field so as not to break backwards compatibility.

It might be better to make these fields non-optional, but that would be a breaking change and I don't want to suggest it in this PR.

[tsmathis]

Awesome, thanks @mkhorton. Is there a reason for the minimum pymatgen version? And if you can fix the linting that would be great.

And apologies for the failing workflows for the tests., we haven't configured the enum generation workflow to be compatible with forks. I've ran the tests locally with your patch and everything looks good.

[mkhorton]

No worries about the tests from my side, I did test this myself also.

For the minimum pymatgen version, this is because SymmetryUndeterminedError was only added back in June. Previously this kind of error wasn't handled well in pymatgen either, for example Structure.get_spacegroup_info() would raise an odd KeyError if symmetry wasn't determined because the symmetry dataset dict would be None. But in either case, the minimum pymatgen version set is already an earlier than the versions pinned in the requirements/ files, so it should be ok.

[tsmathis]

Sounds good, just wanted to be sure there's a reason if we're asked in the future.

[mkhorton]

Just ran through black, hopefully that fixes the linter issues. I don't actually have a dev environment set up for emmet right now otherwise, this is a bit of a hot fix :)

[tsmathis]

All good, pre-commit complains a lot.

You happy with this? Can merge whenever you're ready

[mkhorton]

I'm happy with this, thanks Tyler!