rsz: repairSetup: Restrict number of repairs per iteration

[eszpotanski]

This PR changes the way rsz::repairSetup iterates over the violations. Instead of going one by one and trying to completely repair a violation, this PR introduces the maximum number of repairs per iteration.
If a violation was not repaired, it goes back to the queue and algorithm proceeds to the violation with the worst slack.

This approach should minimize the probability of falling into local optimums, and potentially make fixing other violation easier. Base on the experiment, it can speedup the repairs which resolves all violations, from 6% and up to 25%:

Design: ibex, stage: 2_1_floorplan

Strategy	Min [s]	Max [s]	Mean [s]	Median [s]	Relative
baseline	34.07	34.43	34.25 ± 0.15	34.25	125%
Max 5 repairs per iteration	27.40	27.52	27.46 ± 0.04	27.46	100%
Max 10 repairs per iteration	27.42	27.50	27.46 ± 0.03	27.47	100%
Max 20 repairs per iteration	32.26	32.57	32.44 ± 0.12	32.51	118%
Max 30 repairs per iteration	31.64	31.91	31.78 ± 0.10	31.78	116%
Max 40 repairs per iteration	31.67	32.37	31.94 ± 0.25	31.86	116%

Design: ibex, stage 4_1_cts

Strategy	Min [s]	Max [s]	Mean [s]	Median [s]	Relative
baseline	141.94	143.26	142.46 ± 0.58	142.17	110%
Max 5 repairs per iteration	130.77	132.21	131.71 ± 0.66	132.14	102%
Max 10 repairs per iteration	144.99	145.81	145.28 ± 0.38	145.04	112%
Max 20 repairs per iteration	133.92	136.12	135.09 ± 0.91	135.23	105%
Max 30 repairs per iteration	129.04	131.44	129.85 ± 1.12	129.09	100%
Max 40 repairs per iteration	375.26	381.54	377.54 ± 2.84	375.81	291%

Design: black_parrot, stage: 2_1_floorplan

Strategy	Min [s]	Max [s]	Mean [s]	Median [s]	Relative
baseline	54.34	55.46	55.03 ± 0.49	55.29	106%
Max 5 repairs per iteration	51.83	52.31	52.03 ± 0.20	51.95	100%
Max 10 repairs per iteration	52.00	52.89	52.47 ± 0.37	52.52	101%
Max 20 repairs per iteration	52.13	52.73	52.41 ± 0.25	52.36	101%
Max 30 repairs per iteration	52.21	52.64	52.42 ± 0.18	52.40	101%
Max 40 repairs per iteration	52.12	52.75	52.40 ± 0.26	52.32	101%

Design: black_parrot, stage: 4_1_cts

Branch	Min [s]	Max [s]	Mean [s]	Median [s]	Relative
baseline	192.68	194.13	193.40 ± 0.59	193.39	111%
Max 5 repairs per iteration	173.96	174.63	174.31 ± 0.28	174.35	100%
Max 10 repairs per iteration	174.06	175.66	174.80 ± 0.66	174.68	101%
Max 20 repairs per iteration	174.20	174.45	174.30 ± 0.11	174.24	100%
Max 30 repairs per iteration	173.69	174.57	173.98 ± 0.41	173.70	100%
Max 40 repairs per iteration	174.15	175.71	174.76 ± 0.68	174.40	100%

Design: ariane133, stage: 2_1_floorplan

Strategy	Min [s]	Max [s]	Mean [s]	Median [s]	Relative
baseline	89.18	91.54	90.27 ± 0.97	90.08	107%
Max 5 repairs per iteration	83.93	86.98	85.92 ± 1.40	86.84	103%
Max 10 repairs per iteration	84.02	86.44	85.22 ± 0.99	85.18	101%
Max 20 repairs per iteration	85.16	86.72	85.75 ± 0.69	85.36	102%
Max 30 repairs per iteration	83.70	86.34	85.05 ± 1.08	85.12	101%
Max 40 repairs per iteration	81.91	84.82	83.56 ± 1.22	83.94	100%

For stages where all violations are not resolved, it can increase the number of iterations and improve the slack:

Design: aes, stage: 4_1_cts

Strategy	Mean [s]	Median [s]	Relative	Number of iterations	Worst slack	Total negative slack
baseline	69.88 ± 2.89	70.2042	105%	1331	-37.1303	-1127.02
Max 5 repairs per iteration	66.73 ± 1.85	66.6787	100%	917	-41.5148	-1822.57
Max 10 repairs per iteration	72.13 ± 0.88	71.645	107%	1200	-36.0651	-1602.71
Max 20 repairs per iteration	78.81 ± 1.16	78.1942	117%	1400	-33.651	-1020.18
Max 30 repairs per iteration	80.75 ± 2.08	80.0711	120%	1561	-32.8517	-1411.56
Max 40 repairs per iteration	73.42 ± 1.60	73.8805	110%	1433	-29.8058	-1163.81

Design: ibex, stage: 5_1_grt

Strategy	Mean [s]	Median [s]	Relative	Number of iterations	Worst slack	Total negative slack
baseline	116.72 ± 0.05	116.74	103%	493	-67.1365	-588.325
Max 5 repairs per iteration	218.10 ± 4.85	220.85	195%	475	-57.9249	-1036.24
Max 10 repairs per iteration	111.28 ± 3.29	113.16	100%	442	-59.0674	-954.488
Max 20 repairs per iteration	121.88 ± 3.89	121.13	107%	495	-60.2138	-569.195
Max 30 repairs per iteration	119.03 ± 4.51	117.11	103%	499	-67.1365	-620.345
Max 40 repairs per iteration	116.94 ± 4.03	116.25	102%	489	-67.1365	-892.091

As results depends on max number of repairs per iteration, default value was set to 20 (usually ) and -max_passes_per_iter parameters was added for repair_timing to make it adjustable.

[github-actions]

clang-tidy made some suggestions

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"