There is a bug where two CPU-bound normal tasks sharing a CPU are mistakenly boosted by stalld queue_track backend, even though they already receive an equal 50% of CPU time.

Reproducer:

1. Start stalld with the queue_track backend on CPU 0:
   • taskset -c 0 stalld -v -b queue_track
2. Start two normal CPU bound tasks on CPU 1:
   • taskset -c 1 bash -c 'while :; do :; done' &
   • taskset -c 1 bash -c 'while :; do :; done' &
3. Using watch -n 0.1 ps -a -o pid,%cpu,cls,rtprio,cmd, observe that the two processes receive 50% of the CPU, but they are constantly boosted by stalld.
4. Restart stalld and the issue is gone.