[DP-101] Possibly incorrect monitor behavior

[Imported from JIRA. Reported by Facundo Dominguez @facundominguez) as DP-101 on 2015-02-09 20:22:23]
The following program blocks because the expected ProcessMonitorNotification never arrives, but a monitor notification would be expected according to the semantic notes.

Tested with runhaskell monitorTest.hs

import Control.Distributed.Process
import Control.Distributed.Process.Node
import Network.Transport.TCP
import qualified Network.Socket as N

main :: IO ()
main = do
  Right (transport, internals) <- createTransportExposeInternals "127.0.0.1" "8080" defaultTCPParameters
  lnode0 <- newLocalNode transport initRemoteTable
  lnode1 <- newLocalNode transport initRemoteTable
  runProcess lnode0 $ do
    self <- getSelfPid
    liftIO $ runProcess lnode1 $ do
      send self ()
      liftIO $ N.close =<< socketBetween internals (nodeAddress $ localNodeId lnode0)
                                                   (nodeAddress $ localNodeId lnode1)
      send self ()
      -- removing this timeout has the test complete
      receiveTimeout 100000 [] :: Process (Maybe ())
      monitor self
      ProcessMonitorNotification _ _ _ <- expect
      liftIO $ putStrLn "SUCCESS!"

Start by running the provided monitorTest.hs program with runhaskell and compare its behavior with the semantic notes referenced in the issue. Trace the monitor and ProcessMonitorNotification handling involved in the connection close and timeout; done means the expected notification arrives and the program reaches the SUCCESS output.