Expose quad lights with ANARI

[szellmann]

This PR exposes quad lights via the Anari interface, plus a couple of fixes to make quad lights work again in shader code.

I'm currently computing the "emission" value as color*intensity, but I'm not sure how barney interprets the physical quantities involved, exactly. (I opted for intensity as that was directly accessible from within Vela.) It would probably be (more) correct to interpret emission as color*radiance, assuming that quad lights emit the same amount of energy in all directions and at all angles (Cycles, e.g., has a spread parameter that allows to control the angular dependency, which Anari does not have, meaning it's probably more complicated than that). In that case we should probably convert intensity to radiance, but again not sure how barney would interpret that, exactly.

At the moment, for my test scenes I have to set very high intensity values to see something (order of intensity=5000 for a scene where other devices require order of intensity=1.0). I also noticed that Barney applies inverse square fall-off to quad lights in the PDF and reservoir sampling weights, which I think is only valid for point light sources. (Point lights are also way darker than with other devices.)

Anyway, this PR enables quad lights to begin with, and I'm happy to adapt how emission is computed.

[iwald-nvidia]

i'll merge this already, BUT: could you possibly provide some sort of mini-reproducer in pynari for testing? just a plane with a box or sphere, and some quad light(s) shiniing on it?

[iwald-nvidia]

BTW: "I'll already merge this" means "I'll merge this if and when the tests have run through" :-)