bytes: add assembly version of IndexByte for ARM
benchmark old ns/op new ns/op delta
BenchmarkIndexByte32 459 126 -72.55%
BenchmarkIndexByte4K 52404 10939 -79.13%
BenchmarkIndexByte4M
54470800 11177370 -79.48%
BenchmarkIndexByte64M
1010803000 178860500 -82.31%
benchmark old MB/s new MB/s speedup
BenchmarkIndexByte32 69.58 252.63 3.63x
BenchmarkIndexByte4K 78.16 374.42 4.79x
BenchmarkIndexByte4M 77.00 375.25 4.87x
BenchmarkIndexByte64M 66.39 375.20 5.65x
R=rsc, minux.ma
CC=golang-dev
https://golang.org/cl/
6106044