This exposes the count of issues or MRs per value stream stage. To
avoid counting large amount of DB rows, we use a limit count.