Photometry joint fit: refactor to allow user-input reflection laws

[sblunt]

In response to #384 (comment), we should add capability for users to define their own reflection laws (in addition to the Lambertian disk approx currently implemented). It would be good to support two primary modes:

1. a function (like the Lambertain approx)
2. interpolating using a grid of outputs from an atmosphere model