ensure that Bilby can run on large-scale computing resources beyond the LIGO Data Grid

There are an increasing number of dedicated LVC and external shared computing resources available via the Open Science Grid (e.g., dedicated LVC: Virgo computing centers, Georgia Tech Tier-3 cluster's, LSU's SMIC allocation, CIT/LLO/LHO clusters, etc.; and shared: OSG opportunistic cycles, XSEDE allocations, commercial cloud CPUs, etc.). Bilby should be designed to be able to use them.

Concretely this means: not relying on shared filesystems but rather using explicit Condor file i/o; understanding and being explicit about runtime CPU, memory, and disk requirements rather than assuming an unspecified system configuration; and managing runtime software/library/package dependencies via LDG Singularity containers (ideal), CVMFS (good), and/or self-contained packaging/deployment (also good but harder).

@james-clark, @pfc, Josh Willis, and others are available to help consult and debug.