bugfix: max_structs is np.array in padding
What's Changed
* max_structs is np.array in padding
* add models.png into doc by yury-lysogorskiy in https://github.com/ICAMS/grace-tensorpotential/pull/3
* upd docs/requirements.txt (add mkdocs-glightbox) by yury-lysogorskiy in https://github.com/ICAMS/grace-tensorpotential/pull/4
**Full Changelog**: https://github.com/ICAMS/grace-tensorpotential/compare/0.4.2...0.4.3