Described at Dispatching containers to cloud VMs:

• a single attempt is made to send SIGTERM to the container
• container state/priority fields are not affected
• assuming SIGTERM works, the container record will end up with state "Cancelled"

Currently (without this API) an admin can use the controller API to update the container's priority to 0, with similar results. This API is a bit more direct: it works even if the dispatcher isn't syncing with the controller properly/quickly, and it bypasses the TimeoutSignal config that rate-limits SIGTERM attempts per container.