Sloved a socket-error problem , when socket error happened, some cleaning works should be done immediately

Add a cluster-reconnection method. when cluster nodes lost all connections, it will attempt to re-connect to target cluster every 2 mins

Simply add a logger class, with the help of which , develpers can re-direct qDebug(), qWarning(), qCritical() and QFatal(...) outputs to log files , instead of stderr.

Add a auto-join cluster method, as soon as the server node be started, a join-cluster approach will be taken.

There is still a problem in heavily-test.

const QByteArray & type paras, will cause lift cycle problems. using Value-pass instead.
----
QString, QByteArray supports ref-count and auto-delete , also, these classes support
least-copying, deep-copy will only take place, when the data is about to be changed.
so, in singnals and slots, using QString, QByteArray instead of const QString & and const QByteArray & ,
will not cause performance problem.

Each server instance knows exactly how many clients each server currently have.

branch "SMARTLINK" is continous be ok for smartlink project

This svr 2 svr network using zoomPipeline techniques, an Multi-Threads、pipelined dealer.

The former client focus on heavy-payload test, this one focus on function-test.

Add national-ts files.

Start database develop

set how many thread will be created for ssl trans, how many for plain trans,
client use ssl can communicate with plain client as well