Change default rolling update strategy for MachineDeployments

We currently have an absolute setting maxUnavailable=1 for the MachineDeployments rolling update strategy. And we have a conditional maxSurge=0 for CAPM3 and maxSurge=1 for all other cases.

As result, in VM infrastructures, we have maxUnavailable=1 and maxSurge=1

By default for VMs, it would provide a more predictable experience to have maxUnavailable=0 and maxSurge=1 (the default of ClusterAPI by the way)