Investigage slicing by n optimizations

[est31]

There is a way to optimize CRC performance by using various super fast slicing by n optimizations (another link and another one)

An implementation using const fn would require const fn to stabilize... without const fn it produces a bit of ugly code sadly :/

I should investigate in finding the sweet spot between additional memory requirements and speedups. Maybe I'll just wait until const fn stabilizes and create a build time configuration flag.