Retry storms overwhelm recovering backend services

When multiple clients retry simultaneously during service recovery, amplified request volume prevents the backend from recovering and can cause cascading failures.

Reduce retry attempts to 2, increase initialInterval to 500ms minimum, apply circuit breaker middleware after retry middleware, enable rate limiting before retry middleware, and rely on Traefik's built-in jitter for retry timing randomization.