[cassandra] Create a script to cleanly restart the cluster
The process to restart the cluster is to stop each node one by one and wait it's fully restart.
It can take ~15mn per node to restart, so it's time consuming for an operator to monitor to do this process manually.
A draft of a possible script: https://gitlab.softwareheritage.org/infra/sysadm-environment/-/snippets/1493
The last problem remaining to solve is to use a user allowed to connect to each node of the cluster without authentication and allowed to execute sudo systemctl restart cassandra@[instanceX]