It can happen that polling takes quite some time (say, a week) before the final response arrives.
To solve these issues, would be good to release CRLs every time a "waiting" response has been received where the "checkAfter" period is longer than a configurable time, and to fetch them again when doing the ness poll request.
Moreover, the device/system/application doing the CMP request might get restarted in the meantime, so wold be nice if a resumption of the pending cert request activity was possible, which requires persistent storage of status data like the type of request, request ID, transaction ID, and sender nonce.