Make non-source processors invalid if scheduled for Primary Node only

There is never a reason a to schedule a processor to run on Primary Node Only unless the processor is a source processor. However, we see this misused fairly frequently, and it can result in data not moving through the system. We should update the processor validation logic so that a processor is made invalid if has an Execution Node of Primary Node Only and has any incoming connections.

This includes self-looping connections. Otherwise, Node A can be primary node, and then queue up data. Node B then becomes Primary Node, leaving data sitting on Node A indefinitely.