Simulations use real bias/dark frames from the respective cameras. Currently have grabbed a set of 10 r1 biases (from the optics lab?), and chose the biases based on the simulated frame's seqno%10. The flats actually come from the optics lab frames, which are real imaging flats but which have masked edges.

Fine for now, but I'd like use pointers to newer data, and be able to use different sets (when the r1 electronics get replaced in a few weeks, say)