If you see a message in the UI about approaching limits, it’s just a UI bug right now.
That’s good to know! We had been a little worried about that. Still, with over 500 org members and over 800 contributors, we frequently clog the queues with many hours worth of tests. So the interest in not needlessly running them remains.
Do you think it would be feasible to have our bot handle the launch of workflows? Basically, we could have our app listen to the checks events, and once the linters have passed, trigger the workflow via the new API call I believe I saw.