Parent D490

There is also a Power assembly version available for <= POWER 7,
but it was not ported.

I am getting ~10% improovement on sha256 and 40% improovement on sha512.

Before:

|  nanosecs/byte   mebibytes/sec   cycles/byte  auto Mhz

SHA224 | 3.24 ns/B 293.9 MiB/s 6.15 c/B 1895
SHA256 | 3.24 ns/B 294.7 MiB/s 6.11 c/B 1887
SHA384 | 2.69 ns/B 354.8 MiB/s 5.10 c/B 1895
SHA512 | 2.69 ns/B 354.7 MiB/s 5.10 c/B 1895

After:

|  nanosecs/byte   mebibytes/sec   cycles/byte  auto Mhz

SHA224 | 2.91 ns/B 327.4 MiB/s 5.52 c/B 1895
SHA256 | 2.92 ns/B 326.1 MiB/s 5.54 c/B 1895
SHA384 | 1.82 ns/B 523.4 MiB/s 3.45 c/B 1895
SHA512 | 1.82 ns/B 525.0 MiB/s 3.44 c/B 1895

Fixes T4530

This is integrated into the existing tests, and they all pass. I modified the benchmark.c program to benchmark the hashes, but I wasn't happy with the code quality enough to submit that.