Added a comment explaining why unprioritized jobs are not handled for non-drainer locking.

Author: nickolas-pohilets

stdlib/public/Concurrency/Actor.cpp

@@ -1665,6 +1665,11 @@ retry:;
 
     // Claim incoming jobs when obtaining lock as a drainer, to save one
     // round of atomic load and compare-exchange.
+    // This is not useful when obtaining lock for assuming thread during actor
+    // switching, because arbitrary use code can run between locking and
+    // draining the next job. So we still need to call processIncomingQueue() to
+    // check for higher priority jobs that could have been scheduled in the
+    // meantime. And processing is more efficient when done in larger batches.
     if (asDrainer) {
       newState = newState.withFirstUnprioritisedJob(nullptr);
     }