Playbook for releasing new MacOS instance image to AWS

After the initial build of a production MacOS image with tart and nesting (gitlab-org/ci-cd/shared-runners/images/macstadium/orka#103 (closed)) we need playbook and tooling to regularly release images. It's essential that we can release updates in a timely manner (e.g. for security patches) and that we exercise the mechanism regularly.

Because of the time required to replace an instance (40 minutes scrubbing + 40 minutes startup) we have to either do it very slowly (draining and replacing only a percentage of the fleet at a time) or extend the blue/green runner setup to include blue/green autoscaling groups and underlying dedicated hosts. E.g. scale up, switch, scale down.