Prevent e2e Helm install from re-owning pre-applied CRDs

[Copilot]

The ci / e2e workflow failed during operator install because CRDs were applied via kubectl earlier in the job, then rendered again by the drop chart, triggering Helm ownership validation errors.

• Root cause

  • E2E job mixed two CRD lifecycle paths in one run:
    • manual apply: kubectl apply -f config/crd/bases/
    • chart-managed CRDs from charts/drop templates
  • Helm install aborted on existing CRDs lacking release ownership metadata.

• Change

  • Updated the e2e workflow Helm install command to explicitly disable CRD rendering from the drop chart in this path.
  • This aligns install behavior with the existing pre-apply CRD step and avoids ownership conflicts.

• Workflow delta

  helm install drop charts/drop \
    --namespace drop-system \
    --create-namespace \
    --skip-crds \
    --set crds.install=false \
    --set image.repository=controller \
    --set image.tag=ci