Add UTF8 byte matching optimization

[milseman]

Adds a new instruction to match a sequence of UTF-8 bytes directly, rather than individual match-scalar calls.

[milseman]

Comparing against saved benchmark result origin_main
=== Regressions ======================================================================
- CompilerMessages_All                    87.3ms	85.4ms	1.95ms		2.3%
- CompilerMessages_All_Scalar             78ms	76.2ms	1.84ms		2.4%
- EmailRFCNoMatches_All_Scalar            44.4ms	44.2ms	287µs		0.6%
- ReluctantQuant_Whole                    3.57ms	3.37ms	195µs		5.8%
- SubtractionCCC_All_Scalar               5.9ms	5.78ms	117µs		2.0%
- AnchoredNotFound_All_Scalar             5.85ms	5.76ms	90.9µs		1.6%
- ReluctantQuant_Whole_Scalar             3.57ms	3.49ms	75.3µs		2.2%
- IntersectionCCC_All                     5.47ms	5.42ms	53.2µs		1.0%
- IntersectionCCC_All_Scalar              5.47ms	5.43ms	46.6µs		0.9%
- NotFound_All_Scalar                     2.51ms	2.49ms	22.2µs		0.9%
=== Improvements =====================================================================
- EmailLookahead_All                      18.7ms	21.8ms	-3.11ms		-14.2%
- EmailLookahead_All_Scalar               17.5ms	20.5ms	-3ms		-14.6%
- symDiffCCC_All_Scalar                   16.4ms	17.6ms	-1.15ms		-6.5%
- symDiffCCC_All                          16.6ms	17.5ms	-969µs		-5.5%
- EmojiRegex_All                          78.1ms	79.1ms	-968µs		-1.2%
- EmailLookaheadNoMatches_All             21.1ms	22.1ms	-951µs		-4.3%
- EmailRFC_All                            31.8ms	32.7ms	-900µs		-2.8%
- FSPathsRegex                            4.7ms	5.55ms	-843µs		-15.2%
- DiceRollsInText_All_Scalar              26.8ms	27.6ms	-835µs		-3.0%
- DiceRollsInText_All                     28.3ms	29.1ms	-804µs		-2.8%
- EmojiRegex_All_Scalar                   41.6ms	42.4ms	-789µs		-1.9%
- FSPathsRegex_Scalar                     4.57ms	5.29ms	-721µs		-13.6%
- LiteralSearch_All_Scalar                1.67ms	2.36ms	-693µs		-29.4%
- EmailRFCNoMatches_All                   47.1ms	47.8ms	-686µs		-1.4%
- EmailRFC_All_Scalar                     31.1ms	31.7ms	-682µs		-2.2%
- HangulSyllable_All_Scalar               1.89ms	2.57ms	-678µs		-26.4%
- EmailLookaheadList_Scalar               3.9ms	4.56ms	-656µs		-14.4%
- EmailLookaheadList                      4.03ms	4.68ms	-653µs		-14.0%
- LiteralSearchNotFound_All_Scalar        1.62ms	2.19ms	-574µs		-26.2%
- LiteralSearch_All                       2.64ms	3.18ms	-541µs		-17.0%
- HangulSyllable_All                      2.65ms	3.18ms	-533µs		-16.8%
- LiteralSearchNotFound_All               2.6ms	3ms	-398µs		-13.3%
- HangulSyllable_First_Scalar             774µs	1.03ms	-258µs		-25.0%
- InvertedCCC_All_Scalar                  5.72ms	5.95ms	-230µs		-3.9%
- Words_All                               4.41ms	4.63ms	-223µs		-4.8%
- DiceNotation                            4.1ms	4.32ms	-216µs		-5.0%
- DiceNotation_Scalar                     3.99ms	4.18ms	-186µs		-4.5%
- HangulSyllable_First                    1.24ms	1.41ms	-169µs		-12.0%
- EmailBuiltinCharacterClass_All          9.05ms	9.22ms	-162µs		-1.8%
- EmailLookaheadNoMatches_All_Scalar      19.5ms	19.6ms	-155µs		-0.8%
- InvertedCCC_All                         5.82ms	5.96ms	-144µs		-2.4%
- AnchoredNotFound_First_Scalar           5.84ms	5.98ms	-139µs		-2.3%
- Words_All_Scalar                        4.16ms	4.29ms	-127µs		-3.0%
- MACAddress                              2.32ms	2.44ms	-126µs		-5.2%
- MACAddress_Scalar                       2.26ms	2.37ms	-114µs		-4.8%
- IPv6Address_Scalar                      2.21ms	2.32ms	-107µs		-4.6%
- GraphemeBreakNoCap_All                  1.62ms	1.72ms	-103µs		-6.0%
- AnchoredNotFound_First                  8.09ms	8.19ms	-99.2µs		-1.2%
- GraphemeBreakNoCap_All_Scalar           1.54ms	1.64ms	-96.9µs		-5.9%
- EmailBuiltinCharacterClass_All_Scalar   8.93ms	9.03ms	-96.4µs		-1.1%
- IPv6Address                             2.31ms	2.39ms	-84.7µs		-3.5%
- ReluctantQuantWithTerminal_Whole_Scalar 5.93ms	6.01ms	-82.3µs		-1.4%
- BasicBuiltinCharacterClass_All          2.98ms	3.05ms	-73.6µs		-2.4%
- BasicBuiltinCharacterClass_All_Scalar   2.33ms	2.41ms	-73.5µs		-3.1%
- ReluctantQuantWithTerminal_Whole        5.93ms	5.99ms	-66.7µs		-1.1%
- URLWithWordBoundaries_All               1.76ms	1.82ms	-58µs		-3.2%
- URLWithWordBoundaries_All_Scalar        1.64ms	1.69ms	-52µs		-3.1%
- IPv4Address                             1.96ms	2ms	-43.6µs		-2.2%
- Numbers_All_Scalar                      2.02ms	2.06ms	-40µs		-1.9%
- IPv4Address_Scalar                      1.88ms	1.91ms	-33.6µs		-1.8%
- BasicCCC_All                            2.98ms	3.01ms	-26.5µs		-0.9%
- BasicRangeCCC_All_Scalar                3.09ms	3.11ms	-24.7µs		-0.8%
- Numbers_All                             2.51ms	2.53ms	-23.8µs		-0.9%
- CaseInsensitiveCCC_All                  3.31ms	3.33ms	-23µs		-0.7%
- Lines_All_Scalar                        775µs	796µs	-20.2µs		-2.5%
- Css_All_Scalar                          2.09ms	2.11ms	-16.5µs		-0.8%
- EagarQuantWithTerminal_Whole_Scalar     439µs	454µs	-14.7µs		-3.2%
- EagarQuantWithTerminal_Whole            443µs	457µs	-13.8µs		-3.0%
- URLWithWordBoundaries_All_SimpleWordBoundaries 563µs	573µs	-10.5µs		-1.8%

[milseman]

For FSPathsRegex, prior to this the hottest past was matchScalar, accounting for 32ms (15.9%) of execution. With this, the optimizer uses matchUTF for 6ms (3.4%), which is now pretty far down the list. The reason the benchmark only shows a 15% improvement is because other things like backtracking have bubbled up (but without regression) to dominate the slowness. Those will be the topic of future improvements.

[natecook1000]

I think this could be an assertion, but it's probably fine here.

[natecook1000]

Nice!

[milseman]

@swift-ci please test