Pavel Zbitskiy

@mangoplane could you commit the benchmarks into `crypto_test.go` - that's where other cost evaluation benchmarks for crypto opcodes live - I'll try to repro/replay the notebook. @giuliop how about re-running...

I reproduced the results and got the log-poly model fits my data points perfectly ``` log_A coefficient: 0.0 log_B coefficient: 0.30924400959422704 log_C coefficient: 1.2483621313283038 log_A_log_B coefficient: 0.0 log_A_log_C coefficient: 0.0...

@mangoplane could you merge master in?

Fixed outstanding review comments and merged master in

A current delay between accepting and and relaying is: 1. Hashing + cache access 2. Decoding 3. Signature verification within 2ms [batch](https://github.com/algorand/go-algorand/pull/4621/files#diff-d3741bc12fa9faf0a148f027517dbfde4942668add6ee4ac7b4032543db4b0f5R72) 4. txpool Check + Remember delay I.e. there...

Do not merge - the benchmark showed 6.9k tps vs 7.7k tps on master, so networking code appears to be much slower now.

closed-opened to rerun `licence/cla`

@algonautshant could you re-merge master?