IBL doesn't calculate diffuse irradiance

Currently, rather than baking a irradiance map for light probes, we just reuse the lowest mip levels of the convolved environment map to calculate the diffuse irradiance contribution for materials that use IBL.  This leads to a fairly poor approximation and leads to very flat and dark diffuse contributions from IBL sources.

We should refactor how we handling IBL sources to bake an irradiance cube map for use when calculating the diffuse term for materials using IBL.

See the attached image for an example of the dark and flat diffuse results.