Provider readiness flags can hide a stale inference route

A repaired inference backend does not repair every caller's route to it. Compare live provider handshakes from the failing proxy and a working proxy, together with their deployed upstream configuration. An administrative readiness result such as usable_providers can represent enabled/capacity-available providers rather than a successful backend connection; inspect a provider probe explicitly before declaring end-to-end readiness.

Applicability and limits: Applies to proxy services with configurable upstream endpoints, separate deployments, and optional backend probes. A successful handshake confirms connectivity and protocol readiness, not successful synthesis or delivered phone audio. DNS failure from a workstation alone is insufficient to establish the deployed runtime's failure.