Simulating incomplete detectors
Stemming from a conversation with Brian: we should figure out how we want to implement the simulation of incomplete detectors (e.g., IPND, partial FD). The two obvious options are (1) several different geometries defined and enumerated and (2) simulate a full detector always and throw hits away as needed, based on (for instance) a timestamp for that simulated run.
I advocate the second option, but it means that dead mass is incorrectly modeled for partial detectors. Anyway, this probably warrants some discussion at an upcoming offline meeting, since it bears on a lot of areas (geo, DB, MC generation, relating MC runs to real time spans, bad/good chans table format, ...)