added byte shuffle, CRC32c, carryless multiplication, and xor-rotate

[Logan007]

This pull request adds a byte shuffle shf to hash prospector for 32-bit as well as 64-bit hashing functions.

It relies on SSSE3's pshufb instruction and only works on corresponding hardware which should comprise most recent Intel/AMD CPUs. All related additions to the code are guarded by #ifdefs.

-p-provided patterns can use shf or shf:<perm> where <perm> denotes a permutation of the byte positions. In 32-bit mode, shf:03020100 describes identity, i.e. no change of position, and shf:00010203 equals an endianess changing byte swap. In 64-bit mode (-8), those permutations need to be longer, e.g. shf:0605040302010007 corresponding to an 8-bit left rotate. A sole shf employs a randomly generated permutation.

Some additional thoughts especially on 32-bit byte shuffle and their implementation can be found here.

Fixes #7. Fixes #14. Fixes #17.