fix running tests

fix a bug in lure counting algorithm that returned 0 for the number of lures.

add benchmarks, diagrams, timing monitors, and improve performance for nb_gm_001, nb_gm_002, and nb_gm_003.

Fix nb_gm_003 and improve it with chunk-by-chunk sequence generation. It's now non-skewed and number of matches is controlled for a sequence.

Implement nb_gm_001, nb_gm_002, nb_gm_003, nb_gm_004, and benchmarks for all of them plus progressive randomizer