Updates for Vasprun with MD simulations

[gpetretto]

Summary

I propose a few minor changes in the Vasprun object when parsing MD simulations and especially those produced with the ML option.

• the converged_ionic property was False for an MD simulation. But the check did not seems reasonable for an MD
• the number of steps for a simulation that involves ML MD is not given by the length of ionic_steps
• in case of NPT simulation the parsing of stresses for ML MD simulations was missing

[janosh]

The changes sound good to me! Thanks a lot

[janosh]

Can you compress?

[janosh]

Actually, there's already tests/files/vasprun.xml.ml_md. Is that file incompatible with this test?

[gpetretto]

In principle the ml_md file will not cover the case of a simple MD, so for example it should miss this line:

pymatgen/pymatgen/io/vasp/outputs.py

Line 712 in dc2f6ed

return self.nionic_steps

I can remove it if you prefer saving space in the repository.

[janosh]

If the new file covers a superset of test cases, you can remove the old one and plug in the path to the new one instead (only 1 test needs updating).

[gpetretto]

no, the new one will not cover the ML MD, that produces the md_data attribute. The data are partially overlapping, so will not cover the same set of lines. If I need to remove one, I think that removing the vasprun.xml.md would be better.

[janosh]

no need. where you load the file

vasp_run = Vasprun(f"{TEST_FILES_DIR}/vasprun.xml.md.gz")

could you add a brief comment what differentiates this file from vasprun.xml.ml_md to leave a paper trail of our discussion in the code.