Initial provider implementation

[bastjan]

This PR contains the initial implementation of the provider for Cloudscale.

The provider has been observed to work until the Provisioned state on a local non-OCP test cluster, and to the Running state with node linking on our internal OCP lab, without any manual intervention. The steps for this are in the readme under Testing on a.... The CSRs for the new nodes are approved automatically, and a deployment for the upstream node-linker can be created with hack/deploy-nodelink-controller.sh.

The critical code is in pkg/machine/actuator.go.
controllers/machineset_controller.go adds some annotations to the MachineSet to tell autoscaler vCPUs and memory of a new node.

❯ kubectl get machine,node app-zwtl2       
NAME                                     PHASE     TYPE        REGION   ZONE   AGE
machine.machine.openshift.io/app-zwtl2   Running   flex-16-4   rma      rma1   9m1s

NAME             STATUS   ROLES        AGE     VERSION
node/app-zwtl2   Ready    app,worker   4m34s   v1.28.13+2ca1a23

Checklist

• Categorize the PR by setting a good title and adding one of the labels:
  bug, enhancement, documentation, change, breaking, dependency
  as they show up in the changelog
• Update tests.

[simu]

Didn't fully check, but it feels like the flavour/flavor spelling isn't 100% identical across the code base, error messages and comments.

Overall looks good to me, the one thing that needs to be fixed is the flipped CPU/memory extraction from the cloudscale flavor string when annotating the machineset (cf. inline comments).

Note that I didn't review all the tests in detail.