Fix ray obb degenerate axes

[byjtew]

Description

IntersectionTests::rayOBBParametric divides each half axis by its length to recover the box rotation. A zero-length axis (flat box) produces NaN and skewed half axes produce a non-orthonormal matrix, so the ray is silently rejected or the caller crashes. Height sampling hits this on tilesets with such boxes.

Now a single zero-length axis is rebuilt from the cross product of the other two, two or more return no hit, and the transformed ray direction is normalized so skewed boxes are still tested (with the hit distance scaled back to the original ray).

Issue number or link

I can create one if you want.

Author checklist

• I have submitted a Contributor License Agreement (only needed once).
• I have done a full self-review of my code.
• I have updated CHANGES.md with a short summary of my change (for user-facing changes).
• I have added or updated unit tests to ensure consistent code coverage as necessary.
• I have updated the documentation as necessary. <-- internal fix, up to you guys to decide if that requires doc update

Testing plan

New IntersectionTests::rayOBB with malformed half axes test case: flat box hit on its face, flat box missed beside its face, box with no extent never hit, skewed box hit through its center. The test commit fails on main and passes with the fix.

[byjtew]

For context I discovered that one trying to sample altitude on a freshly updated to v1.1 local tileset.
Seems like the cesium-tools could use a fix as well to trim degenerate volumes.