[perf] tokenize some regexp patterns

[atuchin-m]

The PR tokenizes some common regex filters like ||dlscord.*$all (with token dlscord). This reduces the number of filers that always checked (stored with token = 0) that improves the filter matching performance.

-13% on my machine:

rule-match-browserlike/brave-list
master:  time:    [1.5573 s 1.5734 s 1.5927 s]
this PR:  time:   [1.3547 s 1.3618 s 1.3694 s]

[github-actions]

Rust Benchmark

Benchmark suite	Current: 8684dbc	Previous: 33c03d9	Ratio
rule-match-browserlike/brave-list	1963619323 ns/iter (± 12330527)	2083435514 ns/iter (± 5325633)	0.94
rule-match-first-request/brave-list	1118875 ns/iter (± 10421)	1125686 ns/iter (± 7090)	0.99
blocker_new/brave-list	174628523 ns/iter (± 1212720)	182730408 ns/iter (± 2195358)	0.96
blocker_new/brave-list-deserialize	24469477 ns/iter (± 1527515)	22093509 ns/iter (± 288739)	1.11
memory-usage/brave-list-initial	10212216 ns/iter (± 3)	10212224 ns/iter (± 3)	1.00
memory-usage/brave-list-initial/max	62258783 ns/iter (± 3)	62256231 ns/iter (± 3)	1.00
memory-usage/brave-list-initial/alloc-count	1362674 ns/iter (± 3)	1362293 ns/iter (± 3)	1.00
memory-usage/brave-list-1000-requests	2666974 ns/iter (± 3)	2666926 ns/iter (± 3)	1.00
memory-usage/brave-list-1000-requests/alloc-count	71393 ns/iter (± 3)	71337 ns/iter (± 3)	1.00
url_cosmetic_resources/brave-list	190111 ns/iter (± 716)	197629 ns/iter (± 523)	0.96
cosmetic-class-id-match/brave-list	3333567 ns/iter (± 920199)	3419655 ns/iter (± 911093)	0.97

This comment was automatically generated by workflow using github-action-benchmark.

[atuchin-m]

We don't need to update ADBLOCK_RUST_DAT_VERSION, because the matching logic isn't changed. Tt's only a optimization how we store rules. The current .dat files can be used as is.