The `read_vs_write_current_requests` metric histogram is adjusted, in the case of using API Priority and Fairness instead of max-in-flight, to divide by the relevant limit: sum of queue capacities for waiting requests, sum of seat limits for executing requests. ([#110164](https://github.com/kubernetes/kubernetes/pull/110164), [@MikeSpreitzer](https://github.com/MikeSpreitzer))