Enable more refined supercell settings and adapt DFT settings in phonon workflow

[JaGeo]

• Allow for refined supercell settings
• Enforce consistent k-point settings
• Fix linting/documentation
• Fix interface (maybe parameters instead of a dict?)
• Potentially switch off custodian at even more points (e.g., isolated atoms computation: the large sigma handler should not be used there, for example)
• Make the job naming clearer.

• Figure out why the execution leads to a lot of remote errors (our personal setup or some layered flows?)
• remove manual npars until the bug in atomate2 is solved or leave a default NPAR? (BUG: INCAR Updates not working materialsproject/pymatgen#4051)
• switch to latest atomate2 version instead of branch
• check all VASP settings (see if we can switch off some additional files: chgcar etc)
• Additional ideas: test workflow with 1 supercell per structure to check if VASP settings and supercell sizes are okay
• Documentation, linting for test workflow
• Double-check all supercell matrix choices! Did I convert the matrix to the right formats for phonopy and pymatgen?
• Check naming conventions of flows and jobs (similar to Naming conventions in the RSS flows #101)
• add mp id to the naming of jobs?
• displacement adapted in the ml potential runs
• robust fitting parameters for the whole workflow as a default --> new issue
• allow users to adapt VASP settings more easily (could be transferred into another issue) --> docs will be adjusted in docs PR
• fix write benchmark
• check rmse
• deactivate handlers yes/no
• rename jsons for default fitting parameters
• added default soap_dict
• set VASP makers to force_overwrite ? -->new issue
• add RMSE unit test
• add unit tests for phonon_displacement_maker, phonon_bulk_relax_maker, phonon_static_energy_maker, rattled_bulk_relax_maker, isolated_atom_input_set_generator
• add scale_cell unit test

[JaGeo]

@QuantumChemist i will do some more tests myself but you are invited to test the new implementations as well.

[JaGeo]

I would like to highlight one point in the documentation that would help me to run a workflow:

• add more documentation on the choice of the fitting parameter for the phonon workflow. maybe, provide the Si choice as a default or explain how to chose it

[JaGeo]

One additional point:
I think we are currently a bit inconsistent in the choice of parameters (parameter of the class vs. parameter of the make). I would make everything part of the make that could change depending on the input. (maybe, I am not getting it - this is also an option, of course!)

[QuantumChemist]

One additional point: I think we are currently a bit inconsistent in the choice of parameters (parameter of the class vs. parameter of the make). I would make everything part of the make that could change depending on the input. (maybe, I am not getting it - this is also an option, of course!)

yes, this is indeed inconsitent because of how the code grew and the changes made (adding mlip_hyper variable, then removing it etc.). Once the RSS PR is fully done, I can unify it.

[JaGeo]

@QuantumChemist thank you. 😃 Step by step. It is a huge code base!

[QuantumChemist]

@JaGeo , maybe this is a start to keep track of the right formats for phonopy and pymatgen ?