Use GetAttribute to enable Draco quantization when possible

[ebogo1]

This is a temporary workaround for #9847. See google/draco#742 (comment) for more info. Since SkipAttributeTransform breaks lookups with GetAttributeByUniqueId, this PR uses the alternative approach with GetAttributeId and GetAttribute to preserve quantization when possible.

The downside is that this might not work when there are multiple sets of an attribute, i.e. TEXCOORD_0 and TEXCOORD_1. Because of this, we only call SkipAttributeTransform with the new GetAttribute approach for "single" attributes. Otherwise, we continue without quantization and use the GetAttributeByUniqueId call as before.

I'm working on adding back some of the specs that were removed previously, but the core change is all in decodeDraco.js.

[cesium-concierge]

Thanks for the pull request @ebogo1!

• ❕ There was an error checking the CLA! If this is your first contribution, please send in a Contributor License Agreement.
  • Maintainers, this was the error I ran into while attempting to process the CLA check. Please resolve it to continue CLA checking.
  • You'll need to manually check for a submitted CLA

  Error: The service is currently unavailable.
  

• ❔ CHANGES.md was not updated.
  • If this change updates the public API in any way, please add a bullet point to CHANGES.md.
• ❔ Unit tests were not updated.
  • Make sure you've updated tests to reflect your changes, added tests for any new code, and ran the code coverage tool.

Reviewers, don't forget to make sure that:

• Cesium Viewer works.
• Works in 2D/CV.

[lilleyse]

glTF does allow for TEXCOORD_2, TEXCOORD_3, etc so some of this glTF attribute name -> draco name mapping could be organized in a more general way. Though to be fair, TEXCOORD_2 and higher are only used in ModelExperimental and if we're expecting a bug fix in draco soon then maybe we don't have to worry about it here.

[ebogo1]

My thinking was that if there were more than one TEXCOORD, the numbering would go in order, meaning that at least TEXCOORD_1 would exist - in which case, we would not call SkipAttributeTransform(). Is it possible to just have one TEXCOORD and call it TEXCOORD_2? I'm not sure I'm following your comment.

[lilleyse]

Oh that's true, never mind then. From the glTF spec:

All indices for indexed attribute semantics MUST start with 0 and be consecutive positive integers

[lilleyse]

Thanks for the quick turnaround @ebogo1. For 99% of cases we can use quantization again!

[lilleyse]

Forgot to mention, I tested all the 3D models and 3D Tiles related sandcastles and ran the unit tests locally.

[ebogo1]

Thanks @lilleyse. I tested microcosm-draco.gltf.zip (converted from https://github.com/CesiumGS/cesium/tree/main/Specs/Data/Models/GltfLoader/Microcosm/glTF with gltf-pipeline), which has multiple TEXCOORD attributes and all seems to work as expected.