Holistically review the workflow in this repo and standardise on Kubernetes

[masih]

Revise the way this repo works: Change it such that the entire setup is runnable, consistently so, on AWS and bare metal. And there is an audit trail of every change to runtime. This is a problem solved well by orchestrators such as K8S and K8S controllers.

• Reduce the tools used to ideally 2 but no more than 3.
• Set up CD pipeline triggered by PR merges only
• Standardise node setup, network setup to simple K8S manifest deployment that use concretely versioned containers.
• Make every change to infra auditable by git commits.