Add more functionality into LayerStructure.

[davidorme]

Description

[ETA] This PR has been through a couple of incarnations - this description has been updated to reflect the final state of the PR.

This PR implements the updates to the LayerStructure object described in #451 and a few more things.

• The bulky __post_init__ method has been modularised to make it easier to follow.
• The use of the role_indices dictionary attribute has switched over to using properties for a cleaner and more compact API. The underlying _role_indices_bool and _role_indices_int remain but have index_role_name properties to access role indices.
• The filled_canopy, filled_atmosphere and flux_layers roles have been added.
• The new set_filled_canopy method takes an array of canopy heights and updates those roles as well as populating the lowest_canopy_filled attribute for each grid cell.
• Renames some attributes so it is clearer which are counts of layers (canopy_layers -> n_canopy_layers etc).

This will break the current use of layer_structure.role_indices in #421 but that should be a fairly straight swap and results in a much clearer interface: the code explicitly states what roles are needed.

Fixes #451

Type of change

• New feature (non-breaking change which adds functionality)
• Optimization (back-end change that speeds up the code)
• Bug fix (non-breaking change which fixes an issue)

Key checklist

• Make sure you've run the pre-commit checks: $ pre-commit run -a
• All tests pass: $ poetry run pytest

Further checks

• Code is commented, particularly in hard-to-understand areas
• Tests added that prove fix is effective or that feature works

[davidorme]

@alexdewar A bit difficult to review this stuff because it's all such a big jigsaw of things being assembled, but all of the changes we're working on are based around the upgraded LayerStructure class in virtual_ecosystem.core.core_components. That is a centralised tool, shared among all models for communicating and coordinating information about the vertical layering of the model.

It's been a bit of a two stage process - the class was updated and then we've identified how best to build out the class further in rolling out those updates. I think this is where the functionality stops for now, though, and since its so central, it would be really useful to get a focussed review of that element as part of this PR.

[vgro]

LGTM!

[dalonsoa]

Ok, I review the wrong PR despite @davidorme being explicit about which one I should review...

Anyway, these changes make total sense if aggregate roles are really a thing. Nevertheless, there are a couple of suggestions in the other PR that you might want to consider.

[davidorme]

@dalonsoa Aggregate roles are very much a thing but we could just try and define all of them as named roles (like all_soil). I just wanted to avoid future models having to tinker with LayerStructure to work. I did consider adding a requires_aggregate_index attribute to the model to set them all up at the start, but didn't want to further complicate the base model structure.

[dalonsoa]

@dalonsoa Aggregate roles are very much a thing but we could just try and define all of them as named roles (like all_soil). I just wanted to avoid future models having to tinker with LayerStructure to work. I did consider adding a requires_aggregate_index attribute to the model to set them all up at the start, but didn't want to further complicate the base model structure.

No, I think this is the right place, and the way you have implemented it is very neat and flexible. I've just added a small suggestion, but totally optional.

[dalonsoa]

This looks much cleaner 👍