Wait for volumes to be stable prior to upgrade longorn

As discussed in !4173 (comment 2461950223), and explained in longhorn documentation, we should:

Avoid upgrading when volumes are in the “Faulted” status. If all the replicas are deemed unusable, they may be deleted and data may be permanently lost (if no usable backups exist).

While working on longhorn upgrade MR, we faced a suspicious failure in this job: https://gitlab.com/sylva-projects/sylva-core/-/jobs/9756140380 It could potentially be caused by the fact that we upgraded longhorn whereas it had not finished to sync the replicas on the nodes that were just reinstalled during the rolling upgrade.