I observed that, after reading the message sometimes the log level was
wrong. Most likely because it was updated after the message was read.
Now the log level is bound to the message. This also reduces the number
ctk variables.

Fix missing object name in the example.

It allows you to check how the controller behaves and to monitor the changes in doocs histories, for instance.

solves #70

- worked around gcc bug buy adding a useless default constructor
- removed dangerous convenience referene which would be broken by copy constructor and assignment operator

Merge VirtualModules which have the same name. This doesn't really make a difference, but e.g. the output of dump() will be much cleaner.

Fixes #68

- removed sleep
- put reading to end of loop (process initial value before blocking)

added testHideThis. The test does not fails for CMake, but the printout shows that the connections in the CS are wrong

TestableMode: ApplicationModule guarantees that the main loop thread had the testableModeLock at least once. Solves #66

Moved setting the initial version number for all ApplicationModules from Application::initialise() to Application::run(), so the version numbers of the persisted values from the control system are smaller.

  (first part of #65)

THis is necessary for newer Boost (1.71) which puts the exported targets
in Boost_LIBRARIES. If we export that verbatim it causes link issues on
projects using the config scripts from Make

Sometimes the test failed because Input variables were not
yet written. I added some more stepApplication calls in between.
Seems to work now. Although it is unclear to me why sometimes
I can add an stepApplication after writing a variable and sometimes
I can not.