Modeling non-Lambertian flux (using BRDFs)

[sampotter]

We should be able to extend this library to handle arbitrary BRDFs (although it makes sense to focus on important special cases). This is a thread to collect ideas related to this.

Here's a quick to-do list:

• Pick a shape model and a small region of interest (~1K-10K triangles---note that 10K tris => ~1T triples of points!).
• Pick a suitable BRDF.
• Try visualizing the irradiance & compare w/ Lambertian to see the difference.