feat: UseMultilineReverseSuffix strategy for 3.5-5.7x speedup (#97)

[kolkov]

Summary

• New 18th strategy UseMultilineReverseSuffix for multiline suffix patterns like (?m)^/.*\.php
• 3.5-5.7x faster than stdlib on large inputs (was 24% slower)
• 12-130x faster on no-match cases due to efficient suffix prefilter rejection

Benchmark Results (0.5MB log file, 10,000 lines)

Operation	coregex	stdlib	Speedup
IsMatch	20.6 µs	72.2 µs	3.5x
Find	15.3 µs	68.7 µs	4.5x
CountAll (200 matches)	2.56 ms	14.6 ms	5.7x

Algorithm

1. Suffix prefilter finds .php candidates
2. Backward scan to line start (\n or pos 0)
3. Forward PikeVM verification from line start

Files Changed

• meta/reverse_suffix_multiline.go (NEW) - 234 lines
• meta/reverse_suffix_multiline_test.go (NEW) - tests
• meta/strategy.go - new strategy constant
• meta/compile.go, meta/engine.go, meta/find.go, meta/ismatch.go - integration
• CHANGELOG.md, README.md, ROADMAP.md - documentation

Test Plan

• All existing tests pass
• New multiline tests pass
• Correctness verified against stdlib
• Benchmarks show speedup on large inputs
• Linter: 0 issues

Fixes #97

[github-actions]

Benchmark Comparison

Comparing main → PR #98

Summary: geomean 253.8n 252.3n -0.58%

⚠️ Potential regressions detected:

geomean                               ³                +0.00%               ³
geomean                               ³                +0.00%               ³
geomean                         ³                +0.00%               ³
geomean                         ³                +0.00%               ³
MatchAnchoredLiteral/short_match-4                      6.867n ± ∞ ¹    7.191n ± ∞ ¹    +4.72% (p=0.008 n=5)
MatchAnchoredLiteral/no_match_suffix-4                  4.096n ± ∞ ¹    4.363n ± ∞ ¹    +6.52% (p=0.008 n=5)
AnchoredLiteralVsStdlib/coregex_short-4                 8.720n ± ∞ ¹    9.094n ± ∞ ¹    +4.29% (p=0.008 n=5)
ASCIIOptimization_Issue79/medium_WithoutASCII-4         1.713µ ± ∞ ¹    1.784µ ± ∞ ¹    +4.14% (p=0.008 n=5)
BranchDispatch_Coregex/Digits-4                         9.038n ± ∞ ¹    9.655n ± ∞ ¹    +6.83% (p=0.008 n=5)
BranchDispatch_Coregex/UUID-4                           7.057n ± ∞ ¹    7.490n ± ∞ ¹    +6.14% (p=0.008 n=5)

Full results available in workflow artifacts. CI runners have ~10-20% variance.
For accurate benchmarks, run locally: ./scripts/bench.sh --compare