Materials not being cloned correctly

[cdata]

In #421 we introduced an updated cloning routine for glTFs, and merged it with our previous behavior of cloning all materials in the scene.

Unfortunately, I introduced a regression that causes materials not to be cloned in the majority of circumstances, resulting in a lot of bleed-over related to lighting configuration across different copies of the same model.