Today, an attempt to run an aggregation pipeline with a $shardedDataDistribution stage under a causally consistent pipeline raises an unnamed server error during its execution, together with a hard-to-interpret message.

Since the stage is only meant to collect metrics across shards without snapshot read-like guarantee, such a request may be rejected as soon as it gets received - returning more appropriate messaging to the user.