opti: Stop is too slow about core

[jschwinger233]

Core calls docker api to stop containers concurrently just the way core removes containers.
The relative longer waiting for stop is because dockerd implements the ContainerStop by sending SIGTERM followed by SIGKILL after some graceful termination interval (10s by default).

[jasonjoo2010]

Core calls docker api to stop containers concurrently just the way core removes containers.
The relative longer waiting for stop is because dockerd implements the ContainerStop by sending SIGTERM followed by SIGKILL after some graceful termination interval (10s by default).

After checking the codes related to control workloads, yes, they are running concurrently not like how it's done in removing(based on step).
So maybe it just cannot be more efficient stopping the workloads.