Proposed title of this feature request

Allow smaller worker nodes with Red Hat OpenShift Service on AWS with Hosted Control Planes

What is the nature and description of the request?

ROSA HCP changes the nature of ROSA by reducing the infrastructure that customers pay for as part of using ROSA down to just worker nodes. And given that worker nodes do not run large portions of the OpenShift infrastructure like control plane and infra nodes do, we should be able to expand the available instance types to include the smaller and cheaper ones (such as the *.large ones).

Why does the customer need this? (List the business requirements here)

This allows for customers to optimize their cost model for workloads.

For example, semi-persistent workers used for workload infrastructure (ingress controllers, etc.) can use something like x6a.xlarge for the instance types, and the machine pool can be set up to schedule only those things there and these workloads don't need to scale up and down as much.

But then customer workloads (especially bursty workloads prone to scaling up and down) could set up auto-scaling machine sets with t3a.large instances. The cost of *.large instances are roughly 40%~50% cheaper than their *.xlarge counterparts, and the customer can reduce workload packing issues by scaling by pod to machine sets.

Fundamentally, it can encourage cloud-centric horizontal scaling behavior appropriate with ROSA with the right information and guidance.

List any affected packages or components

OpenShift Cluster Manager, OpenShift Machine API Operator