chore: switch to faster pymatgen

[ltalirz]

Graph generation is sped up by orders of magnitude.

Test suite runtime goes down to ~1 minute on CI (30s on my macbook).

• Breaking change (fix or feature that would cause existing functionality to change)

[ltalirz]

Example: str_m1_o12004_LTF.cif 4860 atoms

new

 In [1]: from mofchecker import MOFChecker

In [2]: %timeit -r1 -n1 mofchecker = MOFChecker.from
   ...: _cif('tests/structures/str_m1_o12004_LTF.cif
   ...: ')
12.2 s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)

Note: 9s of these 12s are spent on parsing the CIF

old

In [2]: %timeit -r1 -n1 mofchecker = MOFChecker.from
   ...: _cif('tests/structures/str_m1_o12004_LTF.cif
   ...: ')
11min 25s ± 0 ns per loop (mean ± std. dev. of 1 run, 1 loop each)

[ltalirz]

To update for next release after pymatgen v2022.0.14

https://github.com/materialsproject/pymatgen

[kjappelbaum]

@kjappelbaum https://github.com/materialsproject/pymatgen/releases/tag/v2022.0.15

[codecov-commenter]

Codecov Report

Merging #163 (c409a51) into master (ddd5b4a) will not change coverage.
The diff coverage is n/a.

[ltalirz]

Nice, let's get this out

[kjappelbaum]

hm, yes. The CI failure with AttributeError: module 'pyeqeq_eqeq' has no attribute 'ostream_redirect' on macOS seems to be due to our EqEq package

lsmo-epfl/EQeq#19