The goal of this ticket is to improve our current performance testing for sharded collections with millions of chunks, covering scenarios such as:

• Performing a full refresh on a mongos
• Performing an incremental refresh on a mongos
• Performing a full refresh on a shard, populating the filtering information.
• Query being routed to a specific shard.
• setFCV transitions, looking for long-standing ops performed during FCV upgrading / downgrading.
• high number of collections (10k+)
• ???