Fixes for deployment and recommendations

[s4mur4i]

Is your feature request related to a problem? Please describe.
I was evaluating the product, and during testing of deployment had many issues. I went forward and created a pr with all my findings

Describe the solution you'd like
Details about my finding:

1. With helm 3.14.2 There is no purge option. you defined it and when trying to delete it just gave error.

Flags:
      --cascade string       Must be "background", "orphan", or "foreground". Selects the deletion cascading strategy for the dependents. Defaults to background. (default "background")
      --description string   add a custom description
      --dry-run              simulate a uninstall
  -h, --help                 help for uninstall
      --ignore-not-found     Treat "release not found" as a successful uninstall
      --keep-history         remove all associated resources and mark the release as deleted, but retain the release history
      --no-hooks             prevent hooks from running during uninstallation
      --timeout duration     time to wait for any individual Kubernetes operation (like Jobs for hooks) (default 5m0s)
      --wait                 if set, will wait until all the resources are deleted before returning. It will wait for as long as --timeout

2. The ansible deploys a cluster version 1.17.3. I checked workload docs, and a lot of findings related to incorrect usage.

• Deployments: They have no field called serviceName, those are only specific for Statefulsets.
• Clickhouse: the initcounter was on wrong level, it should be under spec not metadata.labels
• Elk: in service you define nodePort and targetPort for same port. you can only have one of them.
• Etcd: with ansible a single node cluster is deployed. The nodeselector has 2 issues that is defined. it cannot have empty value.. either empty string like "" needs to be provided, or a boolean. since its single node cluster, the pod will never be deployable with that label, so it can be removed.
• Ingress: extensions/v1bta1 is not valid, and networking.k8s.io/v1beta1 is the correct one. you can check api-version on cluster and it returns this one.
• Ports, env: there are some indentations, they are not relevant, just was troubleshooting and was trying to understand it. I think a linter is needed, there were different practices used as i saw.
• Ohsu: a port cant have all 4 options defined.. port/targetPort/cotnainerPort/nodePort. I don't really understand logic there.
• Thanos Query: the livenessProbe and readiness probes were on wrong indentation, so they failed with invalid parameter
• ready_probe: same here, the delay and period keys were onuner tcpSocket which is incorrect.

3. Regarding Pull images. I am not sure it is working as desired. i did a full pull images, and then when pods were starting they got rate limited by dockerhub, for to many pulls. I created a pull secret, and used pulling with it. I think it is better to use the built in mechanism to do it, then doing pulls and pushing to node. I understand the need for local built images, but i just wanted to bring up cluster and use it.

Describe alternatives you've considered
I have no alternatives, just providing feedback about my findings and solutions. They can be pulled in or issue/PR can be closed.