Enable NEP interface

[naik-aakash]

This PR addresses #54 by adding the necessary functions needed for fitting NEP models and testing on github CI (we will mock NEP fits as fitting explicitly requires GPU)

Changes

• Add calorine as dependency in pyproject.toml
• Grouped some installation commands in Docker file
• Added flexibility to MLIPFitMaker (eg:- users can now change labels used for training data; change was needed because nep executable can only read files with .xyz extension and needs xyz file to have specific labels)
• Add nep_fitting function in src/autoplex/fitting/common/utils.py
• Add NEP as option in MLIPFitMaker
• Add NEP as option in phonon and RSS workflows
• Revert GPUMD install in the dockerfile (NEP model training only works with a GPU > thus redundant)
• Simulate calls to nep using pytest fixtures (add mock_nep in conftest.py)
• Add tests
• Resolve merge conflicts
• Address pending review comment
• Update Readme to include GPUMD install instructions link

[QuantumChemist]

Can you wait a bit with this until #280 is done? 😃

[naik-aakash]

Can you wait a bit with this until #280 is done? 😃

Yeah ofcourse, it is not priority now and I already intended to do this after #280 is merged. Thus have only added dependencies no code yet.

[QuantumChemist]

I don't see anything that needs to be improved 😃
Thank you also for making the part with the ref_names and the train/test file more flexible, as well as adding the mock_nep 👍🏻 💟

[QuantumChemist]

Once #333 is merged, this has to be set to None

Suggested change

	static_energy_maker=ForceFieldStaticMaker(
	force_field_name="NEP",
	),
	static_energy_maker=None,

[naik-aakash]

Hi @QuantumChemist, maybe it is better to do it in your #333 itself. I think this PR is now ready to be merged at this point.

[QuantumChemist]

of course, will adjust it in the other PR then 😃

[JaGeo]

@QuantumChemist Please approve if you are happy.

[QuantumChemist]

I just added one optional suggestion.

[QuantumChemist]

Only an optional suggestion to keep the docstrings more consistent:

Suggested change

	Use the glue.xml core potential instead of fitting 2b terms.
	glue_file_path: str
	Name of the glue.xml file path.
	gpu_identifier_indices: list[int]
	List of GPU indices to be used for fitting. Only used for NEP fitting.
	Use the glue.xml core potential instead of fitting 2b terms. Only used for GAP fitting
	glue_file_path: str
	Name of the glue.xml file path. Only used for GAP fitting
	gpu_identifier_indices: list[int]
	List of GPU indices to be used for fitting. Only used for NEP fitting.

and the same for autodelta (which I cannot include in the suggestion feature)

[QuantumChemist]

didnt see the automerge enabled ^^