If memory serves 2013 hardfork was more a "dependency" bug imho. Client with ver <= 0.8.0 cant's store block bigger than 512KB due to a Berkeley db default misconfiguration. It could have happened with any other external tool used by the client.
The bug wasn't nearly so deterministic. Whether or not a particular block would trigger the bug depended on the runtime history of a particular node (when they started downloading blocks, how many orphans they had observed, etc) meaning that binary-identical versions of Bitcoin Core running on machines with identical hardware could exhibit divergent behaviour based on them connecting to different peers at different times.
In that regard what do you think about the plan to isolate the consensus code in a separete library?
Would have been a good idea to do from the beginning. I'm not sure if the existing code can be cleaned up well enough to make a meaningful difference.