Have to ability to start a single node EventStore in complete read-only mode. This will allow the rebuilding of corrupted indexes on a different server from a production backup without worrying about surpassing the production checkpoint due to newly created system events. The rebuilt index can then be restored to the production environment, just catching up the index on the latest chunks on start-up.
Would read-only replica do almost the same? Yes, it will keep getting replicated data, but it is still read-only, so any operation on its index will be using production data without the node having any chance to become a leader.