> rule of zero

A Communicator is now passed to IndexRearrangementForStreams (Commit #7e1ef552). In the constructor of this Communicator MPI_Init is called. Therefore, an additional call to MPI_Init is not needed anymore.

Some variables were not assigned values in the constructor.  This caused an "MPI_ERR_COMM: invalid communicator" error when using multiple gpus.

automatically set para->useStreams to false for single gpu

(cherry picked from commit f931849c)

added WallModelProbe, replaced PostProcessingVariable with Statistic in Probe (and all derived classes) and added some documentation

Added SG model to StressBC, added Timer-class which wraps existing timer in Simulation, added round-off correction in TurbulentViscosityCumulantK17CompChim_Device (testing ongoing)