docs: add experiment tracking integration guide for operators

[zhoward-1]

Summary

• Adds docs/operator-guides/integrations/experiment-tracking.md
• Covers the full operator workflow: verify network reachability, create a ConfigMap with the tracking server URI, inject it into task pods via extraEnv, handle credentials via Kubernetes Secrets
• Includes a multi-cluster section for environments with multiple registered compute clusters
• Draws an explicit boundary between operator configuration and user task code
• Links to related guides: compute cluster registration, worker config reference, model registry integration

Why

The operator guide's purpose is to help adopters integrate Michelangelo with their existing ML infrastructure. Experiment tracking is one of the first integration points teams ask about — this guide gives operators a concrete, server-agnostic path to wire it up without any Michelangelo-specific tracking client or built-in integration.

Test plan

• Verify all internal links resolve (platform-setup.md#worker-configuration, ../jobs/register-a-compute-cluster-to-michelangelo-control-plane.md, model-registry.md)
• Confirm NetworkPolicy YAML is valid
• Confirm ConfigMap and Secret YAML is valid
• Review that no specific tracking server vendor is named or implied

🤖 Generated with Claude Code