refactoring: Rethink how we organize cluster configuration for better maintenability

This is intended to make implementing and maintaining the upcoming node pools feature eaiser.

Reorganizing structs and funcs like this would result in:

* Less conflicts between commits from multiple branches
* Better reusability when we try to make kube-aws provision a separate cfn stack for each part of a cluster e.g. etcd, control plane, worker nodes(incoming as the node pools feature)

Author: mumoshu