tefALREADY

[thejollyrogers]

It is my understanding that tefALREADY should be returned when trying to submit a transaction that is already being applied to a ledger. From discussions with @MonsieurNicolas , this is flaky. This brings me to the first issue:

1. We should guarantee that tefALREADY will always be returned in the case of a duplicate transaction submission to a node that is already trying to apply the transaction.

Secondly, if a duplicate transaction (hash) is being submitted and the transaction is already in a closed ledger, right now stellard returns tefPAST_SEQ. IMO tefPAST_SEQ should be reserved for submitted transactions that have the same seq number as an already submitted transaction but a different hash. That brings me to my second issue:

2. Return a different error than tefPAST_SEQ if we're trying to submit a duplicate transaction that has already been included in a closed ledger.