fixes #410

The problem was that when the nullifiers were generated and persisted in state_nullifiers, they were padded to 32 bytes (because they are big numbers, they can be less than 32 bytes), but in the processing of the events for spent states, the nullifiers were NOT padded. This is not a problem when most of the time the big number from the Poseidon hash ends up being 32 bytes, but is a problem when the hash is less than 32 bytes, as it causes the mismatch when comparing the nullifier ID vs. the spend ID.

The fix is to always pad the state IDs, for both states and nullifiers when processing events