Improve Multi-Region Active-Active profile

[falko]

• [x] switch to internal load balancer
• [x] clean up python script
• [x] rename zone to region in python script
• [ ] show automation and pure kubectl/helm commands side by side in tables
• [x] JSON logging
• [ ] fix zbctl
• [ ] check that cluster in expected state, e.g. using jq on zbctl --insecure status --output json
• [ ] fix make target to pause exporters, which needs curl
• [ ] make bucket optional or move to other profile
• [ ] Add decision table
• [x] remove image: latest from camunda-values.yaml
• [ ] set Helm Chart version explicitly
• [x] Switch back to official helm chart in README and region Makefiles
• [x] generate the right number of initial contact points in Python, e.g. by using clusterSize and regions as input
• [ ] Maybe add workaround for https://github.com/camunda/camunda-platform-helm/issues/1247
• [ ] automate creation of firewall rules
• [ ] automate cutting the connections by
  • [ ] disabling the firewall rule
  • [ ] disabling DNS Chaining
  • [ ] if needed, restarting the Zeebe pods
• [ ] "disable" Elasticsearch exporter for the failed region by routing exporters to a throw-away ES instance in the failover namespace
  • [ ] in surviving brokers
  • [x] in failOver brokers
  • [ ] in failBack brokers
• [ ] enable Elasticsearch exporter for restored region, i.e. throw away the throw-away ES instance and switch back to the original exporter configuration
• [ ] make ES exporter names unique to prevent accidental reuse of remaining sequence data in case of a very fast recovery
• [x] fix #82
• [x] split configuration into common and per-region part
• [ ] asymmetric configuration for webapps
  • [ ] Run Operate & Tasklist in importer-only mode
  • [ ] and/or disable ingress
• [x] Add metrics to k8s
• [ ] use --context on all kubectl calls
  • [ ] explicitly switch kubectl context before running helm