Add spherical shell geometry support to DEM components

[schunkes]

Description

This PR updates DEM-related components to allow using a spherical shell geometry with the DEMExperiment class.

Checklist

• The code follows the relevant coding guidelines
• The code generates no new warnings
• The code is appropriately documented
• The code is tested to prove its function
• The feature branch is rebased on the current state of the main branch
• I updated the change log if relevant
• I give permission that the Eradiate project may redistribute my contributions under the terms of its license

[leroyvn]

Hi @schunkes, this is a first very quick review. Key points:

• Please make the scope of this PR clear and as limited as possible, i.e. stick to extending DEMExperiment with spherical shell geometry support.
• Please review the whole DEM processing code and vectorize it, it will be clearer, more compact and much faster.

[leroyvn]

Hi @schunkes, thanks for proceeding with vectorization. It looks like a round of cleanup is required prior to proceeding with fine tuning, notably paying attention to:

• unit handling;
• docstring and annotation agreement.

Looking again at this, I think we are in a complicated situation where file buffering fundamentally breaks unit support (we already worked around this but it's not bulletproof).

Also, passing through a file write is something we've been trying to avoid and could be delegated to the user without losing too much convenience. The pattern would then be something like this (still breaking unit support):

# Tesselate DEM
dem_mesh: mi.Mesh = make_dem_from(some_data)
# Either use it directly, 
dem_surface = DEMSurface(shape=dem_mesh, ...)
# Or through a file
dem_mesh.write_ply("dem.ply")
dem_surface = DEMSurface(shape="dem.ply")

Let's review your code live after you're done cleaning it up a bit.

[schunkes]

@leroyvn would you mind, taking another look at the code now? I updated it according to what we discussed. Additionally, I gave the BufferMesh a write_ply method. If you want, I can also show you the workflow as it is now. :)

I did not update the tests yet, so do not expect them to work or pass.

[leroyvn]

Hi @schunkes, we're getting there on the design side. There are a lot of details to fix. I've been specific, but I probably missed some. Please:

• Check defaults and type annotations. There are many missing or inconsistent entries. You also sometimes use mutables as defaults, which is very dangerous.
• Check unit requirement consistency in all your functions. To keep it simple and safe, either all args should have units, or none (and then you might have to document arg units).
• Please compile and check the documentation.

[schunkes]

All things we discussed are implemented. I think this can be reviewed again.

[leroyvn]

I applied a round of fixes. Main changes are:

• Fixed and updated documentation (docstrings and Sphinx docs).
• Improved conversion protocols (I added a new to_mi_scalar_transform()converter for convenience).
• Exposed missing members.
• Fixed mutable defaults.
• Converted dict() to literals for consistency.
• Refactored tests a bit to reduce boilerplate.
• Updated change log.

[schunkes]

Looks good :)