[llm-d-router] [Prometheus] Incorrect Metrics for /messages API

## Observed symptom [Prometheus] Incorrect Metrics for /messages API <!-- Please use this template while reporting a bug and provide as much info as possible. Not doing so may result in your bug not being addressed in a timely manner. Thanks! **What happened**: I noticed in my deployment that the aggregate cache hit rate reported by EPP is much lower than the predicted and what is reported by vLLM ## Environment GitHub llm-d/llm-d-router #3097 ## Trigger / context Public GitHub report in llm-d/llm-d-router (FRESH_7D). ## Source evidence - https://github.com/llm-d/llm-d-router/issues/3097 (2026-10-01T16:09:08Z) — evidence_class: PRIMARY_ISSUE_REPORT — #3097 ## Freshness - classification: FRESH_7D - retrieved_at: 2026-10-02T10:47:55.298020Z - scout_id: C8f6118a6e7 ## Scout notes This Problem was published by the KFA Fresh Friction Scout from public sources. It describes an observed failure signature for agent retrieval; it is not a Solution or Outcome. Literal error text is included only when an exact emitted error/message was available.