When optimising the connections (i.e. merging networks), the trigger receivers for triggered networks are now removed from the trigger network. This was previously leading to crashes.

renamed VariableNetworkNode getTriggerReceiver() into VariableNetworkNode getNodeToTrigger() for better clarity

Instead of making the DeviceModule return the same VariableNetworkNode for the same register, an optimisation step before realising the connections (but after all connections have been defined) is performed. In this step, all networks sharing the same (device-type) feeder will be merged. This is more flexible and avoids problems with (potentially different) triggers etc.

- moved the sleep in Application::testableModeLock() so it helps better against false positive stall detections
- added comments to the growingly complex code of Application::testableModeLock()

preparation for future changes: The TestDecoratorRegisterAccessor now gets a ProcessArray as argument, not the more generic NDRegisterAccessor. This allows it to use specific features of the ProcessArray.

detect stalled tests and print a list of variables which contain data not processed by the application

- prevent spamming of some messages in debug testable mode

fixed duplicated triggers when triggering multiple transfers with the same application variable (e.g. as used in tests frequently)

fixed sending triggers multiple times when the trigger is distributed to multiple targets (TriggerFanOut takes care of the distribution now)

made the instance of the Application::testableMode_mutex static, since otherwise the unique_locks returned are no longer valid in any but the first instance of the application (which affects tests)

fixed race condition in testable mode: now stepApplication() will use a semaphore counter to ensure that all processing is done

Added a TestDecoratorRegisterAccessor which is optionally put around accessors to facilitate test of applications. This TestDecoratorRegisterAccessor does not yet do anything, functionality will be added in the next commits.

Rebased the ScalarAccessor and ArrayAccessor classes onto the mtca4u::NDRegisterAccessorBridge class.
Known problems present: code compiles but tests do not run

Changed the FeedingFanOut so that it does not extract the meta-data (name, shape etc.) from the first slave but expects it in the constructor. This is necessary since the meta-data is already given by the application.

fixed variable networks with external trigger not being put into the same TriggerFanOut (which is required since they should be in the same TransferGroup)

- add all networks triggered from the same node to the same TriggerFanOut
- put all feeding nodes in a TriggerFanOut into the same TransferGroup

- some more code cleaning