Handle invalid mesh inertia produced by custom mesh inertia calculator

[iche033]

🦟 Bug fix

Summary

The custom mesh inertia calculator can produce invalid / null inertias, see code in MeshInertiaCalculator. SDFormat currently logs en error but proceeds to return an empty gz::math::Inertiald object with 0 mass and inertia matrix. This sdf error leads gz-sim to shutdown immediately. This PR prevents collision inertial calculation from returning zero inertial values, and instead it will now return default inertia if an invalid inertia is detected.

The other option is to keep existing behavior and return an sdf error but update gz-sim to add a special case for continuing sim with presence of inertial sdf errors, which is also not ideal because 0 mass and inertia matrix could cause weird behavior in physics sim.

I went with the current approach because this behavior is similar to the case when no custom mesh inertial calculator exists, in which case the default inertial values are also returned.

Checklist

• Signed all commits for DCO
• Added tests
• Updated documentation (as needed)
• Updated migration guide (as needed)
• Consider updating Python bindings (if the library has them)
• codecheck passed (See contributing)
• All tests passed (See test coverage)
• While waiting for a review on your PR, please help review another open pull request to support the maintainers

Note to maintainers: Remember to use Squash-Merge and edit the commit message to match the pull request summary while retaining Signed-off-by messages.

[scpeters]

The custom mesh inertia calculator can produce invalid / null inertias, see code in MeshInertiaCalculator.

from looking at MeshInertiaCalculator.hh, I see the phrase "it works on triangle water-tight meshes for simple polyhedron"

for meshes using convex decomposition, I think we could guarantee that the convex hulls of these meshes are water-tight?

[iche033]

I can extend the MeshInertialCalculator to work with convex decomposed meshes as it currently doesn't.

That can be done in parallel. I think we still do need to handle the case that the mesh inertial calculator fails though, i.e. the changes proposed in this PR.

[scpeters]

instead of changing the default behavior, I'd prefer added a policy enum to ParserConfig like in #1543

[iche033]

closing in favor or #1543