inspiral_lr.py: Add population model to ranking statistic. Population model information in inspiral_intrinsics.py.

- pass the SVD bank's PSD to mkwhitened_multirate_src() for use as a phase reference

- allow the zero-latency whitening kernel's phase to be tweaked to approximate the phase response of the kernel corresponding to some reference PSD

- add option to reduce the phase difference between a the FIR filter for the current PSD and the FIR filter for some other reference PSD.  the use case is in reducing the phase discrepancy between a signal and a previously whitened template.

- remove gratuitous [:] slices from expressions
- use .real, .imag array properties instead of numpy.real() numpy.imag()
- move gain & phase calc for zero-latency whitener up a few lines

- can rely on gracedb module import to work

- unused and won't work because services are already advertised

- template bank files are .xml.gz files

- was using "rates" variable with undefined initialization

- the template IDs are stored in the Gamma0 column, which is a float.  the ID values must be ints or the ranking statistic barfs.

gstlal_compute_strain:  added options use either kappa_tst or kappa_pu to compute actuation time delay, instead of handling the timing of the actuation stages individually.

New cmath element lpshiftfreq that reads in a complex stream and multiplies the phase by a constant without affecting the magnitude. This can be useful when a function whose phase is proportional to frequency has a known value at a particular frequency, and the user wishes to evaluate it at a different frequency.

 - bug fixes in lal_adaptivefirfilt
 - Implemented lal_adaptivefirfilt to correct for f_s and Q time-dependence
 - Added options to correct for complex kappas
 - Added EPICS bits to calib state vector

- change the way the name is constructed
- change the properties that are advertised
- based on my current understanding of how this is supposed to work

- based on my current understanding of how this stuff is supposed to work

- replace built-in algebraic model with the actual trigger distribution
  from all of O2, marginalized across the entire bank

gstlal-calibration:  New element lal_adaptivefirfilt that allows constrution of a time-dependent FIR filter with zeros, poles, gain, and a time advance/delay. For details, type gst-inspect-1.0 lal_adaptivefirfilt.