Problem

There is no automated cleanup mechanism. The start_cluster() finally block only closes SSH/SFTP connections — it does not remove containers, volumes, networks, routes, or deployed files. After a failed run, manual cleanup on every node is required.

File: src/initialize.py

Fix

Add a teardown_cluster() method and expose it via CLI flags with the following behavior:

Teardown per node:

1. docker swarm leave --force
2. docker rm -f $(docker ps -aq)
3. docker volume rm $(docker volume ls -q)
4. docker network rm superset-network docker_gwbridge
5. rm -rf /opt/superset-cluster /opt/*.pyc
6. For mgmt nodes: remove keepalived routes and VIP address

Acceptance Criteria

• teardown_cluster() method implemented that removes containers, volumes, networks, routes, and deployed files on all nodes
• Teardown exposed as ./superset-cluster --cleanup --mgmt-nodes ... --mysql-nodes ...
• --redeploy flag triggers teardown then fresh deploy
• Mgmt node teardown removes keepalived routes and VIP address
• Teardown is idempotent (safe to run multiple times)