Use string version of regexp methods to reduce allocs

[Juneezee]

Both (*Regexp).Match and (*Regexp).FindAllSubmatchIndex have string-based equivalents: (*Regexp).MatchString and (*Regexp).FindAllStringSubmatchIndex. We should use the string version to avoid unnecessary []byte conversions.

Benchmark:

var regex = regexp.MustCompile("foo.*")

func BenchmarkMatch(b *testing.B) {
	for i := 0; i < b.N; i++ {
		if match := regex.Match([]byte("foo bar baz")); !match {
			b.Fail()
		}
	}
}

func BenchmarkMatchString(b *testing.B) {
	for i := 0; i < b.N; i++ {
		if match := regex.MatchString("foo bar baz"); !match {
			b.Fail()
		}
	}
}

func BenchmarkFindAllSubmatchIndex(b *testing.B) {
	for i := 0; i < b.N; i++ {
		if match := regex.FindAllSubmatchIndex([]byte("foo bar baz"), -1); len(match) == 0 {
			b.Fail()
		}
	}
}

func BenchmarkFindAllStringSubmatchIndex(b *testing.B) {
	for i := 0; i < b.N; i++ {
		if match := regex.FindAllStringSubmatchIndex("foo bar baz", -1); len(match) == 0 {
			b.Fail()
		}
	}
}

Result:

goos: linux
goarch: amd64
pkg: github.com/johnkerl/miller/pkg/lib
cpu: AMD Ryzen 7 PRO 4750U with Radeon Graphics
BenchmarkMatch-16                         	 2198350	       517.5 ns/op	      16 B/op	       1 allocs/op
BenchmarkMatchString-16                   	 3143605	       371.5 ns/op	       0 B/op	       0 allocs/op
BenchmarkFindAllSubmatchIndex-16          	  921711	      1199 ns/op	     273 B/op	       3 allocs/op
BenchmarkFindAllStringSubmatchIndex-16    	 1212321	       981.0 ns/op	     257 B/op	       2 allocs/op
PASS
coverage: 0.0% of statements
ok  	github.com/johnkerl/miller/pkg/lib	6.576s

[johnkerl]

This is fabulous -- thank you @Juneezee !! :)