Added manager node monitor

[scohen]

Occasionally, our script that's designed to clean up after the project node port doesn't seem to work correctly. Because of this, it makes sense to monitor the manager node for the project node and to shut down when the manager node dies.

[zachallaun]

I've been thinking about this a bit and while it works, I was wondering if there was a lighter-weight solution.

Right now, when we start the remote node, we pass the flag -e "Node.connect(manager_node)". What if, instead, we called something like LXical.RemoteControl.connect_to_manager(manager_node) that was defined like this:

def connect_to_manager(manager) do
  true = Node.connect(manager)

  spawn(fn ->
    Node.monitor(manager, true)

    receive do
      {:nodedown, ^manager} -> System.stop()
    end
  end)
end

Thoughts?

[zachallaun]

I think state is the manager node here, maybe this is worth doing?

Suggested change

	def handle_info({:nodedown, _}, state) do
	def handle_info({:nodedown, manager}, manager) do