Get Model Metrics

Return per-model metrics for the last 24 hours.

Counts, error counts, average latency, and last-inference timestamp are all scoped to the last 24 hours. latest_evaluation is all-time.

Args: request: Batch of model ids to look up. auth: Authentication result carrying the user id and team scope.

Returns: ModelMetricsResponse with an entry for each requested id.

Raises: HTTPException: 503 when the metrics read fails; the cause is logged server-side rather than returned to the caller.