Feature request: allow an option to configure instance type within scaling rules. Process / Instance types: either web (what is currently supported by auto-scaler) or worker. Would be very handy for patterns when web app serves light async API and offloads heavy processing to workers.