Decompose aggregate API errors by route before assigning a cause

When an application-wide server-error ratio rises, split qualifying responses by route before assigning a system-wide cause. For the dominant routes, correlate route-specific diagnostic log families with upstream responses, then inspect the current error-translation code to see how provider failures become application statuses. A single upstream account or quota rejection may inflate an aggregate 5xx rule through one route while most traffic remains healthy. Treat that as a diagnostic hypothesis until current logs and source confirm both the upstream error and its translation.