[FIX] wrong volume and virial sign for left-handed cells

[akwarii]

Summary

Closes #616

• Fix: det(cell) is negative for left-handed cells, so anything derived from it flipped sign. This affected SimState.volume, the NPT integrators and barostats (Langevin, Nose-Hoover, C-Rescale), the cell filters used by FIRE/BFGS/L-BFGS cell optimization, the pair-potential stress and the D3 dispersion stress. Symptoms were a negative volume, flipped cell forces and wrong pressure.

• Add ts.transforms.cell_volume(cell), which returns |det(cell)| for single or batched [..., 3, 3] cells, and use it at every place that computed a volume inline. Previously the abs was repeated at multiple call sites, and some were missed. This helper should help to ensure this problem is not present in future code.

• Switch the volume measurements in the existing tests, and examples/scripts/3_dynamics.py, to the helper.

Tests

• make_left_handed test helper (tests/conftest.py) re-describes the same crystal with a left-handed cell by negating the third lattice vector.
• test_cell_volume_handedness_invariant: helper is positive and shape-correct for single and batched cells.
• test_volume_is_positive_for_left_handed_cell: SimState.volume is positive and equal to the right-handed value.
• test_npt_is_independent_of_cell_handedness: NPT trajectories match between left- and right-handed cells.
• test_cell_optimization_is_independent_of_cell_handedness: cell relaxation matches for the unit and frechet filters, with and without cell forces.
• test_d3_stress_is_independent_of_cell_handedness: D3 stress is unchanged by handedness.

Checklist

• Doc strings have been added in the Google docstring format.
• Run ruff on your code.
• Tests have been added for any new functionality or bug fixes.

[lil-lon]

Thanks for the fix! I'm not a maintainer, but I noticed a few things that could be improved, so I left some comments.

[CompRhys]

Thanks for the fix!