Subgraph prize queries take too long, and commonly run into rate limits and/or failed requests.
Ideally this is solved with a similar approach to 4e0c17c, but may require a caching layer (local or not), estimations, or a different approach to this data.
