Envoy has added a standard access_log_flush_interval feature to access logs. We should remove custom Istio code duplicating this feature and switch to the upstream way - this will help eventual transition to the standard OTLP and telemetry sinks.

A possible issue is that TCP stats sets it to 15s, but the envoy option applies to all sinks at the same time.