Master Scale-up playbook doesn't land the pre-existing service-signer certs

Issue

• When scaling up the cluster (or rebuilding a previously running master) the service signer certs aren't being copied across from on of the pre-existing masters. If one of the [new_masters] being built happens to be the first master in the list of [masters] this can in turn lead to a new CA being generated and inconsistent CAs across the cluster. Otherwise this seems to simply result in the master services failing to start.