Diffuse Color input from metal/rough seems to be too energy conservative in the implementation section

[sebavan]

From this section of the spec: https://github.com/KhronosGroup/glTF/blob/master/specification/2.0/README.md#metal-brdf-and-dielectric-brdf

It looks like we are applying an energy conservation factor to compute the diffuse color whereas it is already embedded into the fresnel term.

Looking into other implementation, I do not seem to be able to find this term anywhere.

cc: @bghgary , @romainguy

[romainguy]

This term appears in the code sample for the BRDF implementation but not in any of the equations. I believe this term shouldn't be there unless I missed something, in which case it should be explained because as it stands it appears out of nowhere. The same term also shows up in the KHR_material_specular extension.

[bghgary]

This most certainly came from me. I've been looking all over my notes and I can't explain why this term is here. My notes (lack of notes)/email suggests that this terms comes from my investigations into Unity, but I can't find this term in the Unity shader sources. I think this term is likely a mistake.

[bghgary]

For reference: #810