How would the runtime know for sure when the container was ready to accept the request? The lambda model--with the container pulling the events instead of them being pushed to it--seems like the correct way to model the concurrency.