Fail startup on inconsistency between metastore and backup

[agavra]

#5741 introduced the ability to backup our metastores onto disk so that we can recover in the case of failure, and the implementation handles inconsistencies between the command-topic and the backed-up metastore. Instead of always handling these issues transparency, we should allow for an option to hard-fail on inconsistencies, which are usually the result of some corrupted command-topic that should be remedied before starting up the server.