Remove all multiplications and divisions from the main decoding loop.

name            old time/op   new time/op   delta
Decode/256-8      323ns ± 0%    293ns ± 0%   -9.29%  (p=0.000 n=5+4)
Decode/1024-8    1.26µs ± 0%   1.14µs ± 0%   -9.48%  (p=0.000 n=6+5)
Decode/4096-8    4.99µs ± 0%   4.51µs ± 0%   -9.55%  (p=0.002 n=6+6)
Decode/16384-8   20.0µs ± 0%   18.1µs ± 0%   -9.54%  (p=0.002 n=6+6)

name            old speed     new speed     delta
Decode/256-8    791MB/s ± 0%  872MB/s ± 0%  +10.34%  (p=0.002 n=6+6)
Decode/1024-8   814MB/s ± 0%  899MB/s ± 0%  +10.48%  (p=0.004 n=6+5)
Decode/4096-8   821MB/s ± 0%  908MB/s ± 0%  +10.55%  (p=0.002 n=6+6)
Decode/16384-8  821MB/s ± 0%  908MB/s ± 0%  +10.54%  (p=0.002 n=6+6)

Change-Id: Ie9f91242ce04c130a77c1184379e3b9de38fe713
Reviewed-on: https://go-review.googlesource.com/c/151199
Run-TryBot: Daniel Martí <mvdan@mvdan.cc>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>