bytes: SSE for bytes.IndexByte on amd64

bytes: SSE for bytes.IndexByte on amd64

Performance on 2.8 GHz Intel Core i7:

Before:
BenchmarkIndexByte4K  1000000              2997 ns/op        1366.70 MB/s
BenchmarkIndexByte4M      500           3049772 ns/op        1375.28 MB/s
BenchmarkIndexByte64M      50          49582280 ns/op        1353.48 MB/s

After:
BenchmarkIndexByte4K 10000000               298 ns/op       13744.97 MB/s
BenchmarkIndexByte4M    10000            285993 ns/op       14665.76 MB/s
BenchmarkIndexByte64M     500           4618172 ns/op       14531.48 MB/s

R=rsc, PeterGo, r2, r
CC=golang-dev
https://golang.org/cl/2888041