take a marginalized likelihood file name in the Data init, simplify some logic involving when to comput quantities needed for likelihood ranking, check for Nans in FAR column before sending gracedb alerts

far.py: support optional instrument lists in several methods used to compute FAP, set FAP, FAR = Nan if the livetime segment is None, None

gstlal_inspiral_marginalize_likelihoods_online: sleep first in the loop, make sure you do not clobber output if the marginalize job doesn't process anything

gstlal_inspiral_marginalize_likelihood: make no files an error only if not allowing skips, add XML metadata, compute the joint backgrounds regardless of what is in the file, only write an output document if input data was processed

gstlal_inspiral_marginalize_likelihoods_online: program to marginalize likelihoods in the online dag in a never ending loop

far.py: simplified min / max likelihood values for joint likelihood background calculation now that the single detector likelihoods are better constrained

sql command files moved to triggers dir where they are used

Added process ids to sngl_burst events, and defined in and out segments such that later programs can remove whitener artifacts.

also implement marginalization over "theta" feature