2.0 models should not have BINORMAL

[lasalvavida]

Currently we write out BINORMAL from COLLADA if it is present, but it is no longer a valid semantic in glTF 2.0 since it can be computed from TANGENT and NORMAL.

[javagl]

This probably refers to https://github.com/KhronosGroup/COLLADA2GLTF/blob/54e9d78169bc06d13430b633ca9352293ec71364/src/COLLADA2GLTFWriter.cpp#L591 ?! (Just asking because it's still called BINORMAL there. But it might not matter, iff this case can simply be omitted...)

[lasalvavida]

It is to be omitted, but thanks for the correction @javagl.