- instead of a default keyword argument of .quantize_horizon_distances(), store it as an instance attribute
- store the value in the XML PDF cache, and retrieve on load
- update pre-computed cache in share/ to new format (= manually add min_ratio Param to XML tree)

- don't gzip files that are already gzip'ed

- remove superfluous array slice operations
- future-proof integer divisions
- "inline" some algebra to reduce swig interaction overhead

- remove several identity math operations
- remove unused "tmp" variable and calculations used to initialize it

- fix calculation of deltaF when integer sample_rate argument is provided

- aesthetics: in appsink_new_buffer(), when histogramming likelihood ratios for zero-lag candidates, don't inline the construction of the instruments tag for the coinc
- in __flush():  when counting non-coincident singles, do not count singles recovered during single-instrument time.  this test is already part of appsink_new_buffer() but was forgotten here
- performance:  in __flush():  when histogramming likelihood ratios for zero-lag candidates, rely on the zero_lag_time_slide_ids set maintained by stream_thinca.last_coincs to identify all-zero offset vectors;  construct the instruments tag inside the conditional to skip when not needed

added idq_aggregator.py to ugly, factored out aggregation code from gstlal_idq_trigger_gen to idq_aggregator.py, removed fixme comments left over from prev patch

gstlal_idq_trigger_gen: fixed issue where last 10s of trigger data before aggregator epoch from dataframes were not being saved, added more error handling

gstlal_idq_trigger_gen: fixed issue with reindexing when not using live data, fixed dataframe-type trigger saving to string issue when moving to new agg dir epoch

gstlal_idq_trigger_gen: added option to read triggers to disk from dataframe (new) or from string, fixed bug for missing triggers 10s after aggregator directory change

fixed bug of not being able to write triggers after aggregator dir change, added better error handling, added fixme and workaround on a bug that causes missing triggers 10s before and after aggregator dir chance

gstlal_idq_trigger_gen: added workaround for hdf writing causing program crash in rare edge cases, added comments, cosmetic changes

gstlal_idq_trigger_gen: fixed issue with dataframe indices not being set correctly, fixed edge case issues in dataframes (nan values, indexing issues) where program would crash

gstlal_idq_trigger_gen: cleaned up idq trigger generation code, fixed bug where triggers up to 1st cadence were not being written to file

gstlal_idq_trigger_gen: added hdf file saving along with trigger aggregation (max by specified cadence)