Issue with AWS service on worker stop/restart

- There are start/stop/restart worker buttons on manage tab for challenge hosts.
- When they stop/restart workers or when we auto-scale worker, something on the backend becomes inconsistent and workers attribute becomes empty and task_def_arn is non-empty. After this, it is not possible to access the worker and I have to manually clear the task_def_arn for that worker.
- We need to explore when this happens and find a good solution for it.