Look into transactions getting stuck in mempool

[spencercorwin]

In the past we've had issues where users report that their transaction is in the mempool seemingly forever. We have to restart the nodes in order to clear the mempool so that the user can try again. This is obviously not sustainable and needs to be looked into.

The way to debug this is to setup a local network with a NEO•ONE node and a C# node and add logging to both. Then relay a transaction from a NEO•ONE client to a NEO•ONE node and see what happens in both the NEO•ONE node and the C# node. Does the C# node receive the transaction? Is it relayed from the N1 node to the C# node? Why is the transaction not cleared from the mempool?