Split techniques for primitives that use different attributes

[lilleyse]

Take into account whether a primitive has normals or texture coordinates when deciding whether to split incompatible techniques/materials.

This targets the 1.0 branch.

[coveralls]

Coverage increased (+0.004%) to 93.982% when pulling 3b34ed4 on separate-techniques-by-attributes into 5048011 on 1.0.

[likangning93]

@lilleyse works as described for input gltf that already has KHR_materials_common, but should this also address generally when a primitive doesn't have enough attributes for the specified material/technique? Models without kmc that run with kmc output also don't hit this code path.

[lilleyse]

should this also address generally when a primitive doesn't have enough attributes for the specified material/technique

I'm not totally sure what the spec would say about that, but it sounds like that model would be invalid.

Models without kmc that run with kmc output also don't hit this code path.

I think by this point a valid glTF would have already been split already, so the same kmc material wouldn't be used by two primitives with different attributes.

[likangning93]

Small thing: I'm reading this spec title as "splitting two materials," like splitting two materials into four materials, instead of something like "splits a material used by meshes with different attributes."

[lilleyse]

Fixed. I used your wording.

[likangning93]

Thanks @lilleyse!