What is the recommended reactor-to-refinitiv host model when building a client side contributor for RCC via the Java-based ETA/Reactor API:
1. Create a single reactor and invoke connect for each channel even though channel A & B are on different Refinitiv hosts?
2. Create a reactor for each channel since they're connecting to different hosts?
The reason for the question is that we're trying to understand some run-time session management issues whereby:
-If we see start to see latency related problems our application is able to seamlessly switch from sending data from one channel (host A) to another channel (host 
-However, if we run into session management issues whereby we receive either a CLOSED_RECOVER stream state (value = 3) or a CLOSED stream state (value = 4); it seems like we are NEVER able to switch flow from one channel to another and the entire reactor has to be shutdown and recreated. Is this expected or are missing something in our implementation of the API?
