Prevent processing nodes in cases where the node ID is not known #80

1. Adding aliases to nodeIdToAlias and aliasToNode ID can result in sending messages to invalid (nextInternallyAssignedNodeID) NodeIDs.
2. createNewRemoteNode may be triggered before the node ID is known
3. @balazsracz says that openmrn has a queue for messages where resolution is not yet complete
   • and the Java implementation has a hashmap with callbacks
4. RemoteNodeProcessor and LocalNodeProcessor are message layer classes, so they should probably only be notified of messages from nodes with known nodeids

• This would prevent feature code (programs using the classes) from trying to send messages to them with invalid (nextInternallyAssignedNodeID) NodeIDs.

I will work on this and test the change, but let me know if you have comments or can explain more about what is trying to be accomplished in case we can accomplish things a different way (such as 3 above).

Removing nextInternallyAssignedNodeID completely would supersede #77

• This may be preferable since the Message layer is only for nodes with addresses, unless sending and the MTI is one of the Global ones.