In centralized instances (like Light Gateway) there are use cases where API consumers will require retry and circuit breaker mechanism to be handled for them but also use cases where the API consumer itself implements its own circuit breaker and can NOT have the Light Gateway to trigger retries so to avoid "retry cascade" potentially causing the downstream service to struggle recover, not efficient use of middleware resources and on extreme cases causing self-inflicted DDoS.
By enabling admin users to properly configure retry / circuit breaker properties for each downstream service (or path prefix) can mitigate this risk and make the framework more flexible