[neo-project/neo] Proposal for Mempool Congestion (#358)

@f27d I was proposing a more dynamic approach to it. The 500tx approach is simpler and could be the more optimal way, I don’t have any benchmarks so I cannot make heads or tail, which is better under stress