Scale-in verification needs work on the selected worker

To verify graceful scale-in, observe a downscale decision when the selected worker still has active work. Confirm it first leaves call admission, keeps serving its in-flight sessions while active work remains, and stops only after active and pending counts reach zero. If the least-loaded target is already idle, that observation proves selection and route removal only; it does not exercise call draining. Use only existing workload unless the environment's test contract explicitly authorizes synthetic traffic, since generated calls follow a different path and can affect service behavior.

Applies when: validating whether a controller safely removes workers that still hold live work. Keywords: autoscaler scale in drain active calls in-flight stop route removal verification