Feature/multithreaded assembly using a coloring scheme

• Multithreading backends and a default backend selection mechanism

• Backend can also manually selected by the user by setting DUMUX_MULTITHREADING_BACKEND via CMake

• Type of selected backend is reported by CMake on configure

• Backend can also be changed "interactively" through ccmake by editing the cache

• Add a simple generic Dumux::parallelFor with one overload so far fully parallelizable loops over integer ranges

• Implements coloring for CCTPfa and Box to enable parallel assembly over colored subsets

• Enables parallel assembly per default if coloring is supported for disc method and a parallel backend is found

• DUMUX_NUM_THREADS can control the number of threads when TBB or OpenMP or Kokkos is the backend

• Support for backends TBB, OpenMP, Cpp (STL paralell algorithms), Kokkos, Serial (strictly serial execution)

• Bugfix in LoggingParamerTree which was not thread-safe (although the const interfaces implied so)

• Use Dumux::initialize in all tests which enables the DUMUX_NUM_THREADS control parameter

• Update examples

• Python bindings always use the Serial backend for now (the current implementation of the wrapper code is not thread-safe)