velero needlessly requires sh in kubectl image

[xnox]

Describe the problem/challenge you have

Currently velero helm chart requires "sh" "which" in kubectl image, which are copied to a shared volume, which is then copied to velero image and expected to be executable there. This is the case with statically linked "sh" and "kubectl" but is not the case with shared linked ones.

https://github.com/vmware-tanzu/helm-charts/blob/8930c4099fa76b14a4edacc01ecf0ee0e5536bdf/charts/velero/templates/upgrade-crds/upgrade-crds.yaml#L80C16-L80C61

However this seems quite complicated, and brings in redudant runtime dependencies.

Describe the solution you'd like

Anything that results in not needing "sh" in kubectl and velero containers.

Option 1)
Can output to file flag be added to velero install command? such that one can execute velero image as velero install --crds-only --dry-run -o yaml -O /tmp/velero-crds.yaml and then later use kubectl image to kubectl apply -f /tmp/velero-crds.yaml without needed a sh to pipe between the two commands?

Option 2)
Why is all of that performed in the first place..... Can one instead directly run velero install --crds-only straight away, without --dry-run and without pipe to kubectl? There really is nothing dry-run about this call

Anything else you would like to add:
Currently these helm charts dictate for containers to have more dependencies in them, than strictly needed or provided by some minimalistic / distro-less container types.

Environment:

• helm version (use helm version):
• helm chart version and app version (use helm list -n <YOUR NAMESPACE>):
• Kubernetes version (use kubectl version):
• Kubernetes installer & version:
• Cloud provider or hardware configuration:
• OS (e.g. from /etc/os-release):

[jenting]

Related to #571