Remove lock for etcd machines if underlying machine is deleted

[vivek-koppuru]

If etcd controller is rolling out new nodes for external etcd, there is a deadlock scenario where the lock is waiting for another machine to initialize even though the underlying machine got deleted for a transient issue.

https://github.com/aws/etcdadm-bootstrap-provider/blob/main/internal/locking/locking.go#L58

We need to fix this by checking if the machine exists to be able to remove the lock and have it reconcile normally.

[rahulbabu95]

aws/etcdadm-bootstrap-provider#41 should fix this deadlock issue.