Recycle belvedere to a new production rancher node

Belvedere is not used anymore now the databases were moved to albertina.

We can use the server as a compute node for the production cluster.

• Perform a fresh bullseye install and name the server rancher-node-metal04 (cannot, no idrac)
• Configure a zfs pool with a subset of disks(?)
• Configure a swap dataset partition at least of the size of the memory (754Gi)
• Add the server to the rancher cluster
• Declare the labels to launch workload on it