Implement support for external Kubeconfig for Helm runner

[pkosiec]

Description

Take the Kubeconfig as optional argument for Helm runner. If not provided, use platform context.

• Update Helm runner code
• Update manifests

Reason

To enable installing Helm charts on external Kubernetes clusters.

Use cases

• System User: Install Helm charts on external cluster

[mszostok]

Findings:

• Add opt to see selected dependencies. Now we see the whole rendered action, which is unreadable as it has: 1,8k lines. It will be also good to add an option to set the level of returned tree:

  workflow:
      steps:
      - - name: create-target-cluster
          interface:
          selected-implementation: rocketchat-cluster-create-target-cluster-rke2-deploy
      - - name: rocketchat-install
          template: rocketchat-cluster-rocketchat-install-rocketchat-install

  As a result I executed a wrong action which started provision external cluster instead of use provided one.

• The cluster provisioning manifest have inconsistent location. It's quite misleading to have such structure in our repo.

  • k3d: manifests/implementation/containerization/k3d/deploy.yaml
  • rke: manifests/implementation/aws/containerization/rke2/deploy.yaml

• We need to create an epic for multi-cluster installation, currently this task solved only a very narrow use-case:

  • Implement opt to pass optional TypeInstances
  • Maybe create k8s.job.runner to execute on external cluster to solve problem with not exposed services (e.g. on-prem db)
  • Maybe use external project to schedule argo pods on different clusters: https://admiralty.io/blog/2019/01/17/running-argo-workflows-across-multiple-kubernetes-clusters/ (this is not generic solution, it's good as long as the Argo is a core runner)
    • check also progress on: https://github.com/argoproj/argo-workflows/issues/3523