string-matcher: reduce per-matcher memory

[adisuissa]

Commit Message: string-matcher: reduce per-matcher memory
Additional Description:
This is similar to #37324 but doesn't use virtual functions (uses std::variant instead).
These are the goals for this new StringMatcherImpl class:

1. improve encapsulation and use polymorphism that will make the code a bit more readable.
2. reduce the amount of memory used by the StringMatcher (e.g., an object of type ExactStringMatcher doesn't need to also hold a pointer to a regex object).
3. reduce the amount of memory that is essentially repeated (removing the kept protobuf matcher object that holds redundant information).
4. replace the uses of the matcher() that are needed to find the contents of the string-matcher by introducing the stringRepresentation() method.

Compared to #37324, this version enjoys the inlining compiler optimizations that can be made when virtual tables aren't used. The downside is the use of std::variant, which may require a bit more memory, and results in a more coupled code for the string matchers, but IMHO it improves upon the current implementation.

Risk Level: low - not being used at the moment (but future PRs will).
Testing: duplicated the tests needed
Docs Changes: N/A
Release Notes: N/A
Platform Specific Features: N/A

[repokitteh-read-only]

As a reminder, PRs marked as draft will not be automatically assigned reviewers,
or be handled by maintainer-oncall triage.

Please mark your PR as ready when you want it to be reviewed!

🐱

Caused by: #37782 was opened by adisuissa.

see: more, trace.

[adisuissa]

I've executed the router_config_impl_speed_test, using the following:
bazel build -c opt //test/common/router:config_impl_speed_test --config=libc++
and then executing:
taskset 0x1000 bazel-bin/test/common/router/config_impl_speed_test --benchmark_repetitions=10 --benchmark_out=output.csv --benchmark_out_format=csv

and got the following results - showing only the mean of each 10 runs, and the second column is "diff%" (negative number implies that the new version is better - less time, positive number implies that the original version is better):

name	diff% ( (NEW-ORIGINAL)/ORIGINAL)*100 )	Orig Real time	Orig CPU time	New Real time	New CPU time
bmRouteTableSizeWithPathPrefixMatch/1_mean	-0.665188	903	902	897	896
bmRouteTableSizeWithPathPrefixMatch/2_mean	-0.306748	979	978	976	975
bmRouteTableSizeWithPathPrefixMatch/4_mean	-1.40845	1137	1136	1121	1120
bmRouteTableSizeWithPathPrefixMatch/8_mean	-2.43031	1400	1399	1366	1365
bmRouteTableSizeWithPathPrefixMatch/16_mean	-3.62615	1959	1958	1889	1887
bmRouteTableSizeWithPathPrefixMatch/32_mean	-5.44506	3070	3067	2902	2900
bmRouteTableSizeWithPathPrefixMatch/64_mean	-7.57405	5339	5334	4935	4930
bmRouteTableSizeWithPathPrefixMatch/128_mean	-8.4785	9940	9931	9098	9089
bmRouteTableSizeWithPathPrefixMatch/256_mean	-9.862	19296	19276	17391	17375
bmRouteTableSizeWithPathPrefixMatch/512_mean	-10.0741	37534	37492	33745	33715
bmRouteTableSizeWithPathPrefixMatch/1024_mean	-13.3069	82592	82506	71590	71527
bmRouteTableSizeWithPathPrefixMatch/2048_mean	-10.5528	174332	174086	155859	155715
bmRouteTableSizeWithPathPrefixMatch/4096_mean	-6.591	333067	332787	311163	310853
bmRouteTableSizeWithPathPrefixMatch/8192_mean	-5.59905	635485	634911	599967	599362
bmRouteTableSizeWithPathPrefixMatch/16384_mean	-8.94865	1301248	1299905	1184862	1183581
bmRouteTableSizeWithExactPathMatch/1_mean	-0.775194	904	903	897	896
bmRouteTableSizeWithExactPathMatch/2_mean	-0.817996	979	978	971	970
bmRouteTableSizeWithExactPathMatch/4_mean	-1.67401	1136	1135	1117	1116
bmRouteTableSizeWithExactPathMatch/8_mean	-2.64097	1403	1401	1366	1364
bmRouteTableSizeWithExactPathMatch/16_mean	-3.71501	1968	1965	1893	1892
bmRouteTableSizeWithExactPathMatch/32_mean	-5.85192	3096	3093	2914	2912
bmRouteTableSizeWithExactPathMatch/64_mean	-8.63469	5425	5420	4957	4952
bmRouteTableSizeWithExactPathMatch/128_mean	-9.28869	10076	10066	9140	9131
bmRouteTableSizeWithExactPathMatch/256_mean	-9.07519	19105	19085	17368	17353
bmRouteTableSizeWithExactPathMatch/512_mean	-7.7453	36615	36577	33777	33744
bmRouteTableSizeWithExactPathMatch/1024_mean	-9.8651	79410	79320	71564	71495
bmRouteTableSizeWithExactPathMatch/2048_mean	-7.65469	175864	175644	162343	162199
bmRouteTableSizeWithExactPathMatch/4096_mean	-3.29229	322320	321964	311665	311364
bmRouteTableSizeWithExactPathMatch/8192_mean	-4.08471	628595	627609	602605	601973
bmRouteTableSizeWithExactPathMatch/16384_mean	-6.73461	1270351	1268997	1184974	1183535
bmRouteTableSizeWithRegexMatch/1_mean	0.0958773	1044	1043	1045	1044
bmRouteTableSizeWithRegexMatch/2_mean	0.475059	1264	1263	1270	1269
bmRouteTableSizeWithRegexMatch/4_mean	1.04615	1626	1625	1644	1642
bmRouteTableSizeWithRegexMatch/8_mean	0.121902	2463	2461	2466	2464
bmRouteTableSizeWithRegexMatch/16_mean	-0.735622	4491	4486	4460	4453
bmRouteTableSizeWithRegexMatch/32_mean	-0.597356	7876	7868	7829	7821
bmRouteTableSizeWithRegexMatch/64_mean	-0.135291	14806	14783	14777	14763
bmRouteTableSizeWithRegexMatch/128_mean	-0.0775874	29672	29644	29652	29621
bmRouteTableSizeWithRegexMatch/256_mean	2.34694	58304	58246	59673	59613
bmRouteTableSizeWithRegexMatch/512_mean	-2.26124	182966	182776	178829	178643
bmRouteTableSizeWithRegexMatch/1024_mean	2.46927	391451	391006	401057	400661
bmRouteTableSizeWithRegexMatch/2048_mean	-1.02613	565276	564648	559373	558854
bmRouteTableSizeWithRegexMatch/4096_mean	-0.588469	1104409	1103201	1098546	1096709
bmRouteTableSizeWithRegexMatch/8192_mean	-6.25331	2668977	2666013	2501743	2499299
bmRouteTableSizeWithRegexMatch/16384_mean	-1.14446	8314663	8306593	8220362	8211527
bmRouteTableSizeWithExactMatcherTree/1_mean	1.95412	1178	1177	1201	1200
bmRouteTableSizeWithExactMatcherTree/2_mean	1.95412	1178	1177	1201	1200
bmRouteTableSizeWithExactMatcherTree/4_mean	2.04255	1176	1175	1200	1199
bmRouteTableSizeWithExactMatcherTree/8_mean	2.04255	1176	1175	1200	1199
bmRouteTableSizeWithExactMatcherTree/16_mean	1.34794	1188	1187	1205	1203
bmRouteTableSizeWithExactMatcherTree/32_mean	1.51643	1188	1187	1206	1205
bmRouteTableSizeWithExactMatcherTree/64_mean	1.34794	1188	1187	1204	1203
bmRouteTableSizeWithExactMatcherTree/128_mean	1.43218	1189	1187	1205	1204
bmRouteTableSizeWithExactMatcherTree/256_mean	1.17845	1189	1188	1203	1202
bmRouteTableSizeWithExactMatcherTree/512_mean	1.26263	1189	1188	1204	1203
bmRouteTableSizeWithExactMatcherTree/1024_mean	1.34907	1187	1186	1204	1202
bmRouteTableSizeWithExactMatcherTree/2048_mean	2.02703	1186	1184	1209	1208
bmRouteTableSizeWithExactMatcherTree/4096_mean	1.17944	1188	1187	1202	1201
bmRouteTableSizeWithExactMatcherTree/8192_mean	1.34907	1187	1186	1203	1202
bmRouteTableSizeWithExactMatcherTree/16384_mean	0.674536	1187	1186	1195	1194
bmRouteTableSizeWithPrefixMatcherTree/1_mean	1.26796	1184	1183	1199	1198
bmRouteTableSizeWithPrefixMatcherTree/2_mean	1.35135	1185	1184	1202	1200
bmRouteTableSizeWithPrefixMatcherTree/4_mean	1.86757	1179	1178	1201	1200
bmRouteTableSizeWithPrefixMatcherTree/8_mean	2.03735	1180	1178	1203	1202
bmRouteTableSizeWithPrefixMatcherTree/16_mean	1.86916	1178	1177	1200	1199
bmRouteTableSizeWithPrefixMatcherTree/32_mean	2.03562	1181	1179	1205	1203
bmRouteTableSizeWithPrefixMatcherTree/64_mean	1.7842	1178	1177	1200	1198
bmRouteTableSizeWithPrefixMatcherTree/128_mean	1.01095	1189	1187	1200	1199
bmRouteTableSizeWithPrefixMatcherTree/256_mean	0.670578	1194	1193	1203	1201
bmRouteTableSizeWithPrefixMatcherTree/512_mean	0.671141	1194	1192	1201	1200
bmRouteTableSizeWithPrefixMatcherTree/1024_mean	0.671704	1192	1191	1201	1199
bmRouteTableSizeWithPrefixMatcherTree/2048_mean	0.672269	1191	1190	1199	1198
bmRouteTableSizeWithPrefixMatcherTree/4096_mean	0.925926	1189	1188	1200	1199
bmRouteTableSizeWithPrefixMatcherTree/8192_mean	0.92437	1191	1190	1203	1201
bmRouteTableSizeWithPrefixMatcherTree/16384_mean	0.756303	1191	1190	1200	1199

combined_route_config_impl_speed_test_10_mean.csv

IMHO, and from running a few more tests, I've observed that these tests are somewhat noisy, and there's no clear winner.
All results seems to be within the margin of error, and not much execution time difference is observed.