Set status on bootstrap master node if none set

[joejulian]

When the first node is prepared by wksctl, the ExistingInfraMachine status is not set, but the plan is. When reconciling, the controller will never set that resource ready status, preventing the capi controller from reconciling that machine.

https://github.com/weaveworks/cluster-api-provider-existinginfra/blob/2c614bca2a92b15d587f76c7c1c3969d31f50201/controllers/cluster.weave.works/existinginframachine_controller.go#L527-L530

$ kubectl get machines
NAMESPACE     NAME           PROVIDERID                                                   PHASE
weavek8sops   master-b3d5c   existingInfra://ip-10-0-194-68.us-west-2.compute.internal    Pending
weavek8sops   master-ny94n   existingInfra://ip-10-0-195-96.us-west-2.compute.internal    Pending
weavek8sops   master-vj22r                                                                Pending
weavek8sops   worker-0cnnv   existingInfra://ip-10-0-130-187.us-west-2.compute.internal   Pending
weavek8sops   worker-4wwcm   existingInfra://ip-10-0-131-199.us-west-2.compute.internal   Pending
weavek8sops   worker-ify5a   existingInfra://ip-10-0-128-18.us-west-2.compute.internal    Pending
weavek8sops   worker-y0bh1   existingInfra://ip-10-0-130-46.us-west-2.compute.internal    Pending

$ kubectl get existinginframachines
NAMESPACE     NAME           READY
weavek8sops   master-b3d5c   true
weavek8sops   master-ny94n   true
weavek8sops   master-vj22r   
weavek8sops   worker-0cnnv   true
weavek8sops   worker-4wwcm   true
weavek8sops   worker-ify5a   true
weavek8sops   worker-y0bh1   true

In this example, master-vj22r was the first (master) machine defined in the machines.yaml.