Further optimizations to VerifyWithBuffer

[boreq]

Environment:

goos: linux
goarch: amd64
cpu: AMD Ryzen 7 3700X 8-Core Processor 

Results from master:

BenchmarkVerify-16           633      18817784 ns/op     5285065 B/op      23057 allocs/op

Results from this pull request:

BenchmarkVerify-16           694      16936649 ns/op     2035855 B/op      17710 allocs/op

This pull request further reduces the execution time to 90% of what it is on master and number of allocs to 77% of what it is on master.

[boreq]

It is getting harder to optimize this as at this point 81% of the execution time is something we can do nothing about: signature verification:

This is the bit that we can still optimize if anyone is curious:

[boreq]

I will check this.

[boreq]

Oh no that was my broken code. 7e77d30