When performing a nested `$lookup`, the shard acting as the router for the `$lookup` must use its routing information cache to determine which shards own chunks of the inner collection. This routing information can always be stale.

Currently, some tests rely on the assumption that after creating both the outer and inner collections, the shard hosting the outer collection has up-to-date routing information about the inner collection. However, this assumption will no longer hold true after SERVER-95393.

This ticket aims to track the necessary work to eliminate this assumption—that routing information is always up-to-date for all shards—before implementing SERVER-95393.