- currently nonsense, redundant since addition of sum-of-SNR^2 cut.  the
  latter will be fixed, cleaned up, and made to incorporate this feature

- for some reason inspiral.py has it's own, unmaintained, versions of command line parsing stuff that's in datasource.py.  I don't know what's going on, but the way it is doesn't work with KAGRA and I think this patch fixes it

- --likelihood-file is renamed to --ranking-stat-output to make its function more clear
- --likelihood-file-cache is renamed to --ranking-stat-output-cache to retain the symmetry
- --reference-likelihood-url is renamed to --ranking-stat-input to make its function more clear

- to --ranking-stat-pdf to more correctly describe what this command line option does.
- also switch to using the load_url() function, instead of load_filename(), to anticipate getting this data from a non-disk location in O3.

- switch numerator and denominator PDFs to a rate.BinnedLnPDF + snglcoinc.LnLRDensity() -based implementation
- add trigger rate tracking feature to denominator, taking advantage of new high-speed coincidence rate machinery in snglcoinc
- add signal rate tracking feature to numerator, taking advantage of pre-computed P(instruments) factors in SNR PDF look-up table
- switch ranking statistic distribution to a rate.BinnedLnPDF -based implementation
- remove search_summary tables, attach all livetime tracking to the trigger rate data in the denominator object

this is a work in progress.  some notable features that are still missing:

- --min-log-L cut based candidate rate reduction feature in gstlal_inspiral is disabled
- online mode for gstlal_inspiral is non-functional

- a variety of import related adjustments and fixes

gstlal_ll_inspiral_pipe: bug fix, moved the statment where we get template_mchirp_dict outside the loop over bank_groups

inspiral_pipe.py: allow for linearly varying ht-gate-threshold
based on max mchirp for a given bin

gstlal_ll_inspiral_pipe: comment out url jobs, further split aggregator nodes and fix some condor command logic

line arguments --min-log-L and --min-instruments

- lloidparts is patched to require the PSD FFT length to be evenly divisible by 4, and then 50% of the FFT length is devoted to zero-padding (1/4 of the FFT block on each side)
- because this halves the fraction of the FFT block occupied by the Hann window the PSD resolution is effectively cut in half, so all places where the PSD FFT length is set are doubled to maintain the existing resolution.
- refs #2676