update python interpreter used in building lalsuite, build lalsuite/ligo-scald from master for post_O3 optimized Makefiles

output path

workflow

gstlal_ll_inspiral_pipe: launch gstlal_inspiral_marginalize_likelihoods_online via vanilla universe rather than locally

Builds from lalsuite master, nominally every two hours.

- more remain, but we're getting close to removing glue as a dependency

- was in glue, now is in lalsuite

post_O3/optimized/Makefile.ligosoftware_gcc_gstlal: update build to revert commit with fastpath in lalburst

needed

members of itacacpad are now freed with the appropriate gsl free functions
instead of the standard free function.

gstlal-calibration/tests/check_calibration/Makefile:  git did something wierd to Makefile.  Hopefully this fixed it.

lal_adaptivefirfilt:  fixed issue with using static filter when model is not available.  Also numerous plotting script updates.

This reverts commit fc02e413, reversing
changes made to 66c25663.

lnlrcdf_signal: approximation to decrease calculation - use candidates instead of template bank to find the best chunk to which each injection belongs (provisional) - in calculating the likelihood ratio, a DIST_STATS file of one chunk was used as a representative for each source type to reduce the calculation time

- one more tweak for next python-ligo-lw release.  why does this program still exist!?