Currently, if the `createUnshardedCollectionRandomizeDataShard` fp is enabled, we will apply the random data shard setting even when tracking a collection. This can cause issues where a collection exists already and has data on shard1 (primary shard) and a random data shard shard0 is chosen which effectively causes the data on shard1 to get orphaned since the catalog entry points to shard0. This is only a concern in test suites where this failpoint is enabled. If the failpoint is disabled, only the primary shard (which is the actual data owning shard for an untracked unsharded collection) is chosen for the catalog entry.