[DT-723] fix: Recreate rollout so the backend's RWO volume can transfer

[devajipatil]

What

Sets the backend Deployment's rollout strategy to Recreate instead of the default RollingUpdate, and makes the strategy configurable through backend.strategy.

Why

The backend keeps its sqlite database on a ReadWriteOnce PVC (emc-1-pvc-emc-backend-data). RollingUpdate starts the replacement pod before terminating the outgoing one, so the new pod tries to attach a volume that is still exclusively attached to another node. It never starts:

Failed to load logs: container "edc-management-console" in pod "emc-backend-655f5fcc7d-cc8np"
is waiting to start: ContainerCreating
Reason: BadRequest (400)

Every backend redeploy currently needs a manual scale --replicas=0 / --replicas=1 to get through. A single-replica Deployment holding a RWO volume should not use RollingUpdate.

How

File | Change -- | -- templates/deployment-backend.yaml | {{- with .Values.backend.strategy }} block rendering spec.strategy, matching the existing nodeSelector / affinity pattern (key at column N, nindent N+2) values.yaml | backend.strategy.type: Recreate + a comment explaining why not RollingUpdate Chart.yaml | 3.15.0 → 3.16.0; removed a trailing-whitespace line at EOF

13 insertions, 2 deletions. The Chart.yaml changes are not cosmetic — ct lint fails a modified chart at an unchanged version, and yamllint (ct's lintconf.yaml) treats the trailing whitespace and missing final newline as errors.