feat(core): implement 3D tetrahedra grid generation

[imrn99]

Description

title

Necessary follow-up

• benchmark the routine and see if we can generate beta values and coordinates on the GPU.
• make a figure which details connectivities and raw value generation

[codecov-commenter]

Codecov Report

❌ Patch coverage is 97.75187% with 27 lines in your changes missing coverage. Please review.
✅ Project coverage is 85.44%. Comparing base (c4a05ee) to head (c511bf0).
⚠️ Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
honeycomb-kernels/src/grid_generation/mod.rs	91.47%	15 Missing ⚠️
honeycomb-kernels/src/grid_generation/internals.rs	98.57%	8 Missing ⚠️
honeycomb-kernels/src/grisubal/mod.rs	0.00%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #372      +/-   ##
==========================================
+ Coverage   84.95%   85.44%   +0.48%     
==========================================
  Files          73       75       +2     
  Lines       10238    10574     +336     
==========================================
+ Hits         8698     9035     +337     
+ Misses       1540     1539       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[cedricchevalier19]

Can you explain the algorithm or give a ref?

Is it really something you want to put in core?

[imrn99]

Can you explain the algorithm or give a ref?

I plan on adding a figure to the documentation before merging. The structure is essentially a basic cut of hexahedral cells into 5 tetra each (4 orthocentric, and one regular at the center adjacent to the other four). This component is repeated along each axis (with an alternating symmetry to get a conformal mesh) to build the grid.

Is it really something you want to put in core?

I thought about this too. I think we could have a grid generation module in honeycomb-kernels, but this means:

1. moving already existing grid generation routines (2D quad, tri, and 3D hex)
2. removinng methods builtin to CMapBuilder

1 is ok, 2 is slightly annoying IMO. We could make the new interface a function (e.g. generate_grid) or keep a builder pattern (e.g. turn the descriptor into GridBuilder with a fn build(self) -> CMap method)

[imrn99]

@cedricchevalier19

• I moved grid generation to the honeycomb-kernels
• I added figures to the doc and updated tests

The CI fails due to a docker failure to fetch taplo, same tests are successful in local.

I'm not sure I can make a cleaner PR, both changes are linked

[cedricchevalier19]

Sorry for making you do a lot of changes in this PR but I think it will be clearer on the long run.