crypto/sha256: speed-up for very small blocks

crypto/sha256: speed-up for very small blocks

Similar to https://golang.org/cl/54391, but for sha256
name          old time/op    new time/op    delta
Hash8Bytes-8     209ns ± 1%     191ns ± 1%  -8.65%  (p=0.000 n=10+9)
Hash1K-8        2.49µs ± 1%    2.47µs ± 2%  -0.74%  (p=0.045 n=9+10)
Hash8K-8        18.4µs ± 1%    18.2µs ± 0%  -0.98%  (p=0.009 n=10+10)

name          old speed      new speed      delta
Hash8Bytes-8  38.1MB/s ± 1%  41.8MB/s ± 1%  +9.47%  (p=0.000 n=10+9)
Hash1K-8       412MB/s ± 1%   415MB/s ± 2%    ~     (p=0.051 n=9+10)
Hash8K-8       445MB/s ± 1%   450MB/s ± 0%  +0.98%  (p=0.009 n=10+10)

Change-Id: I50ca80fc28c279fbb758b7c849f67d8c66391eb6
Reviewed-on: https://go-review.googlesource.com/82995
Run-TryBot: Ilya Tocar <ilya.tocar@intel.com>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Filippo Valsorda <hi@filippo.io>