During disagg fresh cluster initialization, there is a window after writing the initial no-op step-up entry but before writing the first checkpoint in which a crash will result in an unrecoverable cluster – see SERVER-134754 for details.

Currently, during fresh cluster startup, the first time we test the configured KMIP server lands squarely in this window. If the KMIP server happens to be down, and we run out of retries (currently, this happens after 15 seconds of backoffs), we crash and hit the above case, creating an unrecoverable cluster. This ticket will mitigate this in two ways:

1. Extend the total backoff so that we can tolerate transient KMIP server outages.
2. Test the configured KMIP server before writing the initial no-op step-up entry and fail early, avoiding the unrecoverable cluster case.