for requested instruments

- allows its operation to be tested more easily, and also re-used if needed

- move upper roll-off to higher LRs and decrease total number of bins

- previously, slices would not include the bin corresponding to the upper
  bound, but this made it impossible to consistently request open-ended
  slices in binnings with infinite upper bounds.  rate.py now includes the
  bin in which the upper bound occurs in the slice interpretation and this
  updates the places here that need fixing

- to account for probability that a signal is partially recovered by noise.
  what's here will need to be thought about and tuned, but this adds the
  feature.

- use snglcoinc.LnLikelihoodRatio.samples() to avoid repeated evaluation of
  parameter probability densities.  reduces time required for sampling by
  about 30%.  no increase of number of samples, just enjoying the speed
  increase.

- the ccdf and pdf plots had stopped getting generated properly following
  recent work.  not sure why, I think maybe there's a bug in matplotlib
  when x data contains infs.  anyway, this patch makes them work again

Kipp Cannon authored 10 years ago and Kipp Cannon committed 10 years ago

- now that things are log likelihoods, some variable names should be updated

while we figure out how to deal with livetime in the dag

- also, don't "optimize" LigolwSegments sent over web interface

- add a triggersegments segment list type in lloidparts.Handler
- replace Data class' .seglists attribute with a reference to the Handler's dictionary of segment list dictionaries
- update code in Data class to refer specifically to the triggersegments segment list