1) Asynchronous queries are "issued" in the order that you've called them, but their completion (and subsequent processing) is dependent on when the queries complete. Each async query is processed on a separate worker thread, and the EntityManager does track all outstanding async requests but doesn't actually queue them. As async requests complete back on the main calling thread they will be processed in the order they've completed.
2) In terms of the pipeline between client and server, there's really no difference between one EntityManager issuing many async requests vs. many EntityManagers each issuing one request. If each EM is connected to the same EntityServer then they will all use a common communication proxy. It's therefore application dependent how you handle this - it may be cleaner, or even more performant, to use separate EMs to load data and then import into a central EM, but it would really depend on your specific needs and require some performance testing. Remember that an EM is not threadsafe, and can only be safely used by one thread at a time.