- teach cmp_nxydumps how to identify and compare gap intervals
- require firbank_test_01.py to pass a cmp_nxydumps comparison

this is working towards placing the documentation in directories that don't
include the point release

introduce an autoconf macro to do the work that takes care of not
overwriting variables

- port to lal's swig bindings
- aesthetics in exception handling

cbc_template_iir.py: fix the bug when freq f have negative value; add property autocorrelation_mask to templates

gstlal_fake_frames: make dot graph output occur for the PLAYING STATE on presence of GST_DEBUG_DUMP_DOT_DIR

gstlal_fake_frames: address review tasks 1) fix dot graph capsfilter 2) add dot graph pipeline output 3) add a review section to the doxygen documentation

there's no right answer here, but when a timeout occurs the latency is used to guess the timestamp of the missing data, and if that timestamp is then used to udpate the latency the result is the latency keeps growing in step with the passage of time and the fake buffers timestamps all end up being identical.  that's arguably the right answer but not helpful.  by not recalculating the latency in the event of a timeout the fake buffers get monotonically increasing timestamps and the latency is left at whatever it was last measured to be before data stopped flowing.