- teach RankingData class to optionally draw noise parameter samples from a
  different CoincParams object than is used to compute likelihood ratios
- in calc_rank_pdfs, when drawing "with zero-lag", continue to compute
  likelihood ratios from the "without zero-lag" version of the CoincParams
  object

share/Makefile.rerank: first commit

- when removing --samples-file option, forgot some related bits of code

- adjust grid settings
- fix plot limits to minimum single detector snr

- waste of time.
- but do retrieve snr_min from coinc param PDFs object

- change to linear scales
- edit title and axis labels to make it clear these are coincs, not singles, and what "SNR" means in this context.

- also give .tight_layout() a try

- no such thing in this pipeline, that's for a different pipeline

- rename functions
- move axes type (log/lin) adjustments into plot building functions
- clean up associated create_plot() calls

- alphabetize
- add golden_ratio constant
- adjust matplotlib.figure import

- in FAPFAR.__init__(), ranks is ordered so don't need to use min() and max() to find min and max.

- implements inverse of .fap_from_rank().
- needed to determine locations of confidence markers on plots

- instantiate Points classes through self rather than by explicit class name, reduces risk of bugs being introduced by accidentally instantiating the wrong one

- use new abs() feature of offsetvector class to compute back-off

- even though we might not have new triggers, we can still form new coincidences from the triggers we already have if the boundary has moved
- lowers latency:  reports coincs at the end of lock stretches promptly instead of holding onto them until the start of the next lock stretch

- move the input loop from gstlal_inspiral_marginalize_likelihood to a function in far.py
- replace input lops in gstlal_inspiral_calc_likelihood, gstlal_inspiral_calc_rank_pdfs, gstlal_inspiral_marginalize_likelihood, gstlal_inspiral_plot_background, gstlal_inspiral_plotsummary, gstlal_inspiral_rate_posterior with calls to this function

- use this in .__init__() to simplify initialization.