The RTT is not working as expected due to two reasons:

1) The application thread invoking the stub method is sending itselfe 
the reqest on the connection, which is potentialy a blocking operation.

If blocked the application thread has no time control to ensure the 
client timeout.

2) In the client Delegate syncronisation is performed to keep a proper
bind status. For that it was a monitor object (bind_sync) used, that
did't offer timeout facilities. The waiting threads queued at this
object and blocked for undefined time.

Enhancements:

1) The critical section is now controlled by the use of a lock. Application-
threads calling Delegate can now wait only for a given time.

2) Signalisation of the bind status is done by condition variables.

3) Connection establishment and the sendig of the request is done by a
parallel thread. The application thread is waiting for the completion 
of the send thread by the use of a future object for only a given time.

The solution seems to work well with the given test application.