Add triangle_math tests and fix Triangle3d::bounding_sphere bug

[aristaeus]

Objective

Adopted #12659.

Resolved the merge conflicts on #12659;

• I merged the triangle_tests added by this PR and by Math primitives cleanup #13020.
• I moved the commented out code from the original PR into a separate test with #[should_panic].

[alice-i-cecile]

Diff on the final commit looks good to me :)

[IQuick143]

I believe that the computation for the degenerate triangle should be done in the same way as for the obtuse one, which will remove a division by zero bug as well as making the code more regular. (Also depending on what optimisations the compiler misses it might make this path faster by avoiding unnecessary math.)

[IQuick143]

I believe that the computation for the degenerate triangle should be done in the same way as for the obtuse one, which will remove a division by zero bug as well as making the code more regular. (Also depending on what optimisations the compiler misses it might make this path faster by avoiding unnecessary math.)

[aristaeus]

I've simplified the code paths as mentioned, and also added is_obtuse() (and is_acute for completeness).

I've just realised that Triangle2d doesn't have is_obtuse, is_acute (added in this PR) or is_degenerate (already exists). Should I copy those methods over for consistency?

[alice-i-cecile]

I've just realised that Triangle2d doesn't have is_obtuse, is_acute (added in this PR) or is_degenerate (already exists). Should I copy those methods over for consistency?

Yes please: these are really nice little utility methods.

[IQuick143]

Overall it's pretty good, only thing I'd change is optimising away the square roots in the actute and obtuse functions.

[IQuick143]

Thanks for the work on this PR!