Plot horizon distance from ranking statistics

This MR will change the horizon distance plotting program in the DAG: instead of creating one horizon distance plot from reference PSDs, the new program create the horizon distance from the ranking statistic for each template bank bin. The main changes that affect the DAG workflow are summarized below:

• gstlal_inspiral_pipe: change plotting program from gstlal_plot_psd_horizon to gstlal_inspiral_plot_rankingstats_horizon (previously called gstlal_inspiral_plot_psd_horizon which does not accurately describe what it actually does, also see miscellaneous changes below).
• inspiral.py: update the horizon_dist_layer for the new plotting program.
• gstlal_inspiral_summary_page: horizon distances plots are put inside a new UI component (report.ImagePagination) so that all the plots will not be fetched at the same time and can be paginated.
• make the above changes to new style DAG workflow as well.

Miscellaneous changes:

• add new plotting module called horizon.py
• update the gstlal_plot_psd_horizon to use horizon.py module,
• renamed gstlal_inspiral_plot_psd_horizon --> gstlal_inspiral_plot_rankingstats_horizon

Note: this MR depends on gstlal-visualisation/ligo-scald!36 (merged), it's better not to merged before that request has been merged.

gstlal-inspiral/python/lloidhandler.py

@@ -1156,13 +1156,13 @@ class LLOIDTracker:
 			# so that the "how many instruments were on test"
 			# is aware of this buffer.
 			snr_min = self.rankingstat.snr_min
-				for instrument in instruments:
-					# FIXME At the moment, empty triggers are added to
-					# inform the "how many instruments were on test", the
-					# correct thing to do is probably to add metadata to
-					# the buffer containing information about which
-					# instruments were on
-					self.rankingstat.denominator.triggerrates[instrument].add_ratebin(list(map(float, buf_seg[instrument])), len([event for event in events if event.snr >= snr_min and event.ifo == instrument and event.search == "signal_model"]))
+			for instrument in instruments:
+				# FIXME At the moment, empty triggers are added to
+				# inform the "how many instruments were on test", the
+				# correct thing to do is probably to add metadata to
+				# the buffer containing information about which
+				# instruments were on
+				self.rankingstat.denominator.triggerrates[instrument].add_ratebin(list(map(float, buf_seg[instrument])), len([event for event in events if event.snr >= snr_min and event.ifo == instrument and event.search == "signal_model"]))
 
 			# FIXME At the moment, empty triggers are added to
 			# inform the "how many instruments were on test", the