workflow-execution: reimplement workflow stop

[diegodelemos]

Currently, the workflow stop feature is implemented using an auxiliary celery task

On a serverless workflows architecture we won't have Celery service per engine, so we have two options:

• Introduce a REANA-Worker (work is already started to have a REANA-Worker) who will use the mentioned task and it will: kill the dependent jobs as it is done right now, and reimplement the part of killing the workflow (revoke(workflow_uuid, terminate=True)) using Kubernetes API instead.
• Since we already depend on Kubernetes, implement stop within the RWE_ itself, and take advantage of Kubernetes lifecycle hooks, using a PreStop hook which would do the clean-up.

Depends on #158.