Well, it's there! It's really basic right now, but I will expand upon this basis.
I've added a new "homogeneous" volume that is similar to the "homogeneous" volume region except it has no bounds. And my first tests seem to show that it just works and not too bad at that.
I need to make a more general "homogeneous" framework that will accept textured inputs (so that you can use measured spectra for example) and also add support for the fresnel property to this volume. When it's done you'll be able to create muddy water, milk, orange juice, ... (this is possible right now but you have to use a glass material for the IOR, not a glass2, and convert measured scattering and absorption to RGB colors).
Volumetric emission is not yet supported.