Works for puts, gets and transactions and
them also within functions.

This solution allows to limit the number of threads
handling client requests in a server waiting for
a response from a given server.

For example, if we configure the system to allow
at max 1/2 of the available threads to handle client requests
busy towards a given server, supposing we have a
4 servers system with MAX_THREADS=8, if at a given point
in time server1 has 4 threads busy waiting to get a
response from server2 and another request comes to server1
that would require to communicate with server2,
the request will be rejected.

This feature could be used in situations when a server is not responding
or is responding slowly in order to limit the number of server
threads (in the rest of servers) to handle client requests that are
busy waiting for an answer from the problematic server.
This way, even if a server is not responding or is being very
slow, there will always be some threads available to handle
client requests in the healthy servers.
As a result, a non-responding server would not eventually
provoke a 100% traffic loss.

In order to implement this feature, a new observer type
has been created (SendReceiveMessageObserver).
Calls to this observer have been added before
a put, get or transaction needs to go to another server
(before any change has been made locally) and
after the response has been received.

With a particular implementation of this observer
(see GenericDunitTest.MySendReceiveMessageObserver)
that keeps track of the number of threads waiting for
an answer to each servers, operations may be rejected
if the limit has been surpassed.

In GenericDUnitTest that observer is set in the
servers and the functionality can be tested
manually by looking at the logs.

Some refactorings and renamings

…d add function to sanctioned data serializables