[benibela]

I am using Bob Jenkin's one_at_a_time hash for that purpose.

It is really simple and works with unaligned data.

But it is not doing well in benchmarks. I wonder if I should use another one