glTF conformance: khronos-RecursiveSkeletons

[elalish]

This test looks almost correct, but there's some baseline shift. Not sure if this is related to one of the other open skeleton bugs (we do use SkeletonUtils to clone everything in MV, so it could be specific to the cloning step). https://modelviewer.dev/fidelity/#khronos-RecursiveSkeletons

[donmccurdy]

Do you see the baseline shift in https://github.com/cx20/gltf-test?

There is the issue of calculating a bounding box correctly for a SkinnedMesh as well; that seems like the more likely cause if the viewport is off center.

[Mugen87]

Related #21507.

The same approach from the PR could be used for SkinnedMesh, too.

[Mugen87]

If the wrong bounding box of the skinned mesh is the root cause, this issue can be merged in #14499.

[elalish]

That does seem likely.

[Mugen87]

Merging into #14499 then 👍 .