Simplify phonon get_supercell_size() and test clean up

[janosh]

• include method (vasp|force_field) in PhononMaker test names
• simplify phonon get_supercell_size() job
• better use pytest structure fixtures to reduce duplication

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 76.64%. Comparing base (256b39a) to head (48c063b).
Report is 5 commits behind head on main.

❗ Current head 48c063b differs from pull request most recent head aa3d1bb. Consider uploading reports for the commit aa3d1bb to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #783      +/-   ##
==========================================
- Coverage   77.03%   76.64%   -0.40%     
==========================================
  Files         122      114       -8     
  Lines        9108     8502     -606     
  Branches     1429     1275     -154     
==========================================
- Hits         7016     6516     -500     
+ Misses       1663     1599      -64     
+ Partials      429      387      -42     

Files	Coverage Δ
src/atomate2/common/jobs/phonons.py	75.28% <100.00%> (+0.01%)	⬆️
src/atomate2/common/schemas/phonons.py	94.07% <ø> (ø)

... and 15 files with indirect coverage changes

[janosh]

waiting on uv release astral-sh/uv#313 (comment)

[janosh]

looks like we can successfully migrate CI deps installation to uv. @utf are you ok with this? should give a decent speed up as in materialsproject/pymatgen#3675 (comment)

[janosh]

uv install time is ~40s vs pip's ~150s, so a 3.7x speed up

[janosh]

looks like the test failures could be related to recent jsanitize changes in materialsvirtuallab/monty#638. @Andrew-S-Rosen do you happen to know a good fix here?

[Andrew-S-Rosen]

@janosh: I must admit, I can't see how that PR would cause the issue here. The command being called here is jsanitize(obj, strict=True, allow_bson=True), but the logic I added only kicks in if recursive_msonable=True (which is not the case by default). I also don't think replacing an isinstance() check with try/except would trigger this error in particular. That said, the timing is suspicious. 😅 There was also materialsproject/maggma#930 in maggma, but that doesn't seem to be called here.

[JaGeo]

I think we already talked about a similar issue here ( #781 , not the CHGNet issue but the one of the band structure workflow). I think emmet was at fault.

[JaGeo]

See #781 (comment)

[janosh]

@Andrew-S-Rosen my bad, should have looked more closely at the stack trace. was going mostly by temporal correlation 😅

@JaGeo thanks for the pointer! i missed that discussion 🙏