Place attempt metrics outside automatic retry adapters

Automatic HTTP retries must sit inside the instrumentation boundary when a metric promises one observation per physical attempt. Wrapping a retrying client call records only the final logical result, hides recovered failures, and folds backoff plus several requests into one latency sample. Either implement retries above the observed call, instrument the transport adapter itself, or explicitly document the metric as logical-operation telemetry. This distinction matters even when the final response is successful.

Applies when: HTTP clients use transparent status or connection retries while dashboards claim per-attempt reliability. Keywords: HTTP retries requests urllib3 adapter observability metrics latency failures