upgrading sylva triggers a workload cluster node rolling update (OS image change)

When we upgrade Sylva for the mgmt cluster, any change of sylva_diskimagebuilder_version propagates to workload clusters via shared-workload-clusters-settings, resulting in those workload clusters doing a node rolling update at their next sylva-units HelmRelease periodic reconciliation.

This is because, we have:

shared_workload_clusters_values:
...
  sylva_diskimagebuilder_version: '{{ .Values.sylva_diskimagebuilder_version }}'

(https://gitlab.com/sylva-projects/sylva-core/blob/f3e60419ebe4b87e902da2ad4d2d1f42718e417f/charts/sylva-units/values.yaml#L6619)

This is not the behavior we desire (workload clusters lifecycle is meant to be decoupled from the lifecycle of the mgmt cluster). This is in fact a remnant of the behavior we had in early Sylva days where we had a single sample workload cluster with a lifecycle fully coupled with the one of the mgmt cluster.