Slightly better benchmarks for when string and separator are equivalent and also less branching in inner loops.
benchmark                        old ns/op    new ns/op    delta
BenchmarkGenericNoMatch               3430         3442   +0.35%
BenchmarkGenericMatch1               23590        22855   -3.12%
BenchmarkGenericMatch2              108031       105025   -2.78%
BenchmarkSingleMaxSkipping            2969         2704   -8.93%
BenchmarkSingleLongSuffixFail         2826         2572   -8.99%
BenchmarkSingleMatch                205268       197832   -3.62%
BenchmarkByteByteNoMatch               987          921   -6.69%
BenchmarkByteByteMatch                2014         1749  -13.16%
BenchmarkByteStringMatch              3083         3050   -1.07%
BenchmarkHTMLEscapeNew                 922          915   -0.76%
BenchmarkHTMLEscapeOld                1654         1570   -5.08%
BenchmarkByteByteReplaces            11897        11556   -2.87%
BenchmarkByteByteMap                  4485         4255   -5.13%
BenchmarkIndexRune                     174          121  -30.46%
BenchmarkIndexRuneFastPath              41           41   -0.24%
BenchmarkIndex                          45           44   -0.22%
BenchmarkMapNoChanges                  433          431   -0.46%
BenchmarkIndexHard1                4015336      3316490  -17.40%
BenchmarkIndexHard2                3976254      3395627  -14.60%
BenchmarkIndexHard3                3973158      3378329  -14.97%
BenchmarkCountHard1                4403549      3448512  -21.69%
BenchmarkCountHard2                4387437      3413059  -22.21%
BenchmarkCountHard3                4403891      3382661  -23.19%
BenchmarkIndexTorture                28354        25864   -8.78%
BenchmarkCountTorture                29625        27463   -7.30%
BenchmarkFields                   38752040     39169840   +1.08%
BenchmarkFieldsFunc               38797765     38888060   +0.23%

benchmark                         old MB/s     new MB/s  speedup
BenchmarkSingleMaxSkipping         3367.07      3697.62    1.10x
BenchmarkSingleLongSuffixFail       354.51       389.47    1.10x
BenchmarkSingleMatch                 73.07        75.82    1.04x
BenchmarkFields                      27.06        26.77    0.99x
BenchmarkFieldsFunc                  27.03        26.96    1.00x

R=dave, fullung, remyoudompheng, rsc
CC=golang-dev
https://golang.org/cl/7350045