Limits queries from one organization (or from one user on single-organization clusters) to run on a certain percentage of mapper threads at most, effectively throttling queries to prevent one organization from consuming all capacity.