Document rough benchmarks in README.md

[alexanderkjeldaas]

Especially for hashing, the performance is one of the main drivers for a particular hash function.

When evaluating which hash function to use, I'd like to know some rough speeds. For example how does the various Blake2* versions relate to SHA-3 and SHA-2 in these particular implementations?

[ocheron]

This is reasonable to ask but I don't think anyone will want to maintain benchmark results in README. We have many compiler versions, flags, target architectures that influence the result.

A first step could be better organized benchmark code to launch easily with cabal bench or stack bench.