It's by design. It's designed to enable doing load shedding, not to massively kill running tasks. What would be the use case where there would be >10 running tasks that must be killed (vs just let go)?

I'll mark as available for now as I don't think this is needed often enough to warrant preempting what Kevin has on his plate.

SGTM. Maybe the template needs to be updated? It's strange that all bugs for swarming UI are auto-assigned to him.

It's useful while running led-triggered experimental jobs on production hardware, e.g. when testing a new builder that schedules many tasks blocking production tasks. It's often clear from the first task that something is wrong and killing the remaining tasks is useful to prevent rare capacity to be wasted.

Ah, so you want to kill a tree of tasks. Filed issue 878895 and duped against it.