According to this What is firestore Reference data type good for?. Currently it looks like you have to do a round trip to fetch the data for each document reference in your original query.
Is this on the roadmap to do this expand/aggregation in the cloud?