Remove scene context from errors/warnings in Jolt Physics module

[mihe]

Fixes #102638.
Supersedes #102726.

Warning

This PR is entirely untested as of writing this.

This PR removes any mention of specific scene nodes in the various error/warning messages found in modules/jolt_physics, due to calls to Object::to_string being problematic when using physics/3d/run_on_separate_thread.

This PR also changes most of these warnings to be WARN_PRINT_ONCE instead, since there's no real context provided anymore, and therefore no point in spamming the warning continuously.

I've also gone ahead and removed most of the errors related to invalid scaling (e.g. scaling a cylinder on anything but its height axis) as previously emitted by the JOLT_ENSURE_SCALE_VALID macro, as they were already adding some friction for some users, and without context they're likely to cause even more friction, especially when emitted during runtime.

Now instead we only do a WARN_PRINT_ONCE inside of JoltShape3D::make_scale_valid if a significant invalid scale is detected, and only for DEV_ENABLED builds. This should allow us to catch these kinds of errors when bugs are reported that suffer from this, but they won't bother end users.

I've left in the errors for zero-scaled transforms, as emitted by the JOLT_ENSURE_SCALE_NOT_ZERO macro, as they will typically cause a lot of problems and thus should be dealt with in some way, which I currently do by just resetting the entire Basis to identity. That seems to me like a significant enough of an intervention to warrant emitting an error, even without any context.

[akien-mga]

This PR removes any mention of specific scene nodes in the various error/warning messages found in modules/jolt_physics, due to calls to Node::to_string being problematic when using physics/3d/run_on_separate_thread.

I missed that the linked issue was only when running physics on a separate thread (which isn't the default). In that case maybe it still makes sense to preserve the detailed messages for the case where physics/3d/run_on_separate_thread is false, and only fall back to a more generic message in the other case?

[mihe]

In that case maybe it still makes sense to preserve the detailed messages for the case where physics/3d/run_on_separate_thread is false, and only fall back to a more generic message in the other case?

Having two different error message everywhere would be quite messy. I could however make it so that any node name is reported as <unknown> or something when running on a separate thread, which wouldn't require changing the existing messages in any way. I already do this if the instance ID (ObjectID in the case of nodes) isn't found in ObjectDB.

[Calinou]

I could however make it so that any node name is reported as <unknown> or something when running on a separate thread, which wouldn't require changing the existing messages in any way.

That makes sense to me, as long as we document this limitation in run_in_separate_thread's documentation.

[mihe]

You'll find this simpler implementation in #102726.

[akien-mga]

Superseded by #102726.