As suggested by @lisandrasilva, we should add the ability to filter out failures happening due to a specific reason, e.g., not report overflows.