Right now, the $listClusterCatalog requires a local read concern to be able to execute the $_internalListCollections stage that runs under the hood.

The problem is that all the following $lookup stages running against the config database also run with a local read concern, meaning the reads on the config catalog may not be consistent.

If the reads of the config catalog are not consistent, we could potentially return wrong information to a user whenever there are DDL concurrent operations.

Some consequences could be to return an empty list of shards for a tracked collection. It could happen, for example, when a resharding operation is running.

This ticket aims to ensure $listClusterCatalog makes consistent reads on the config catalog.