Currently initial sync generates new idents for collections and indexes in the copying phase even if replicated catalog identifiers are enabled. We should instead obtain the idents from the source node and reuse them so that the node's catalog post-initial sync is consistent with the source's. The approach for this proposed in the SPM-4173 design is to add an undocumented option to listCollections and listIndexes which makes it include the ident in the output, and then use that ident. We do not want to publicly expose the existence or values of idents so it should not be included by default.

The oplog application phase of initial sync already replicates idents.