reduce allocs by using optimized single use Keccak256/512 Hash

[jmank88]

This PR aims to speed up and reduce garbage, and allocations in general, from hashing. The main change is optimized forms of Keccak256/512 for one-time usages. Some additional tweaks were made at call sites as well to reduce copying and allocations. These new benchmarks illustrate the difference:

BenchmarkKeccak256/unoptimized-4         	 2000000	       921 ns/op	     960 B/op	       4 allocs/op
BenchmarkKeccak256/optimized-4           	 2000000	       729 ns/op	     448 B/op	       1 allocs/op
BenchmarkKeccak512/unoptimized-4         	 2000000	       884 ns/op	    1024 B/op	       4 allocs/op
BenchmarkKeccak512/optimized-4           	 2000000	       691 ns/op	     512 B/op	       2 allocs/op