[rancher] Add 2 management nodes for the archive-staging-rke2 cluster

Regarding the issue with etcd and ceph latencies (#4876 (closed)), we need to use local hypervisor storage to improve the IOs.

Unfortunately, the sratch storage of the hypervisors is not redounded so we need to add at least 2 etcd nodes on different hypervisors to ensure the cluster status is not lost in case of a disk issue.

It will also avoid cluster issues when maintenance is needed on the management nodes