Define project-wide data models

[Sinclert]

This PR completes the remaining tasks of issue #482: the migration of data model property definition from its current state (array-based) to an object oriented design (class-attributes).

🔍 - Scope

This PR introduces a set of data models that are used project-wide. Thanks to their definition, we could expect less bug-leading syntax like object[0] and more meaningful syntax like object.name.

🌿 - Additional Improvements:

• Typing notations!
• Enums values for Systematic attributes.

📝 - Commits methodology

Each commit within the refactor has a prefix indicating:

• models: refers to the creation of the data models.
• h5py: refers to the adaptation of the HDF5 MadMiner interface functions.
• internal: refers to the code base migration in order to use the newly defined models ✨ .
• tests: refers to the small adaptations that tests needed to suffer after adapting the codebase.

🔄 -Compatibility

⚠️ This PR breaks backward compatibility ⚠️

There are multiple public-methods which change their expected arguments:

• LHEReader.add_cut() and DelphesReader.add_cut() renamed argument pass_if_not_parsed to required.
• LHEReader.add_efficiency() renamed argument value_if_not_parsed to default.
• combine_and_shuffle removed unused argument overwrite_existing_file.

[Sinclert]

NOTE:

This PR contained several bugs, fixed in commits d3e8653 and d9ee1e5.