Event-relayer: do not regenerate the proof if file exists

[sept-en]

Currently, if the relay restarts, it will start from the latest block so it most likely will generate proofs for every transaction until it catches up. Generating proofs for transactions takes most of the time so if we first check for the proof file existence (we already store built proofs) we will significantly reduce the time needed to catch up after the restart.

But also be aware that we need to take into account that file could be corrupted so at least add some check for file size (higher than some threshold)