Publish a multi-cluster observability reference architecture

Report needed documentation

The current observability reference stack covers a single cluster or the control-plane cluster in a split deployment. Operators need a provider-neutral architecture for observing self-managed NVCF across control-plane and compute-plane clusters.

Describe the documentation you'd like

Describe how metrics, logs, and traces are collected, routed, correlated, stored, and queried across clusters.

The architecture should cover:

• Control-plane, compute-plane, node, and workload health.
• Cross-cluster identity and signal correlation.
• Collector and backend placement, availability, and failure behavior.
• Network and security boundaries.
• Retention, storage sizing, and scaling considerations.
• Integration with the existing dashboards and infrastructure-sizing guidance.
• The boundary between the example single-cluster stack and a production multi-cluster deployment.

Start by reading the existing single-cluster observability reference stack, dashboards, and infrastructure-sizing guidance mentioned in the issue. Produce a provider-neutral multi-cluster architecture covering signal collection, routing, correlation, storage, security boundaries, failure behavior, scaling, and the boundary between the example and production deployments.