Accounting for FHI-aims species' defaults

[ansobolev]

Summary

This PR adds new SpeciesDefaults object to account for files with the default definitions of species-related keywords (integration grids, basis sets...) in FHI-aims, named species' defaults. This is done to abstract away the work with species' defaults from the AimsControlIn object, to bring the API more in line with VASP, and to enable running heterogeneous calculations that need different precision levels for different atoms.

Major changes:

• An AimsSpeciesFile class representing the defaults for one species is introduced
• A SpeciesDefaults class representing a list of species' defaults is introduced; the AimsControlIn class now uses SpeciesDefault
• An AIMS_SPECIES_DIR configuration variable is introduced which should point to the root of FHIaims/species_defaults directory
• Breaking change: the possibility to use default species' defaults directory is removed. Every input sets need explicit definition of the species' defaults directory relative to FHIaims/species_defaults or FHIaims/species_defaults/defaults_2020.

Todos

None

Checklist

• Google format doc strings added. Check with ruff.
• Type annotations included. Check with mypy.
• Tests added for new features/fixes.
• If applicable, new classes/functions/modules have duecredit @due.dcite decorators to reference relevant papers by DOI (example)

[tpurcell90]

Would it be possible to reintroduce the default species_dir behavior? For some people using only one set of defaults is relatively okay

[tpurcell90]

otherwise this looks good from an fhi-aims perspective. This will break atomate2 tests but we can fix that once it is merged

[ansobolev]

Would it be possible to reintroduce the default species_dir behavior? For some people using only one set of defaults is relatively okay

Yeah, it could be possible (just remove one check and add species_dir to paths_to_try, I guess), but based on my discussions over here people also want to freely change the defaults, and they don't like writing out the full path every time.

[tpurcell90]

Okay that makes sense

[ansobolev]

Hi, is there anything that should be done here?

[JaGeo]

@ansobolev Thanks for contributing in any case! I alao hope one of the maintainers will soon have some time! 😃

[shyuep]

Thanks for the contribution. I have merged it!