Adds ability for service operators to customize pod tolerations similar
to how resource limits/requests are currently handled.

Features:
- Add Tolerations field to ContainerSpec API type
- Implement merge behavior: custom tolerations are merged with defaults,
  overriding by key when same key exists
- Set global default tolerations (node.kubernetes.io/not-ready and
  node.kubernetes.io/unreachable with 120s timeout) in controller
- Update deployment templates (managers.yaml, operator.yaml) to render
  custom tolerations from Deployment struct
- Add test coverage for merge logic and override behavior

Example usage:
```yaml
operatorOverrides:
- name: "keystone"
  controllerManager:
    tolerations:
    - key: "node.kubernetes.io/not-ready"  # Override default timeout
      operator: "Exists"
      effect: "NoExecute"
      tolerationSeconds: 600
    - key: "node.example.com/gpu"          # Add new toleration
      operator: "Equal"
      value: "nvidia"
      effect: "NoSchedule"
```

The merge behavior ensures operators get both default tolerations
(unless overridden by matching key) plus any additional custom ones,
providing flexibility while maintaining safe defaults.

Jira: OSPRH-18693

Assisted-by: claude-4-sonnet

Signed-off-by: Martin Schuppert <mschuppert@redhat.com>