crc32: add slicing-by-8 algorithm

This patch series adds a new algorithm slicing-by-8 to the existing version of crc32.c and cleans up some minor warnings from sparse. It also adds a functional self test that can be used to both test the correctness and performance of the crc32 algorithm on different architectures.

The main change from the previous version of this patch set is to break up the patch into several smaller hunks.