Fix various msvc compiler warnings. With msvc the compiler flags are explicit passed to nvcc with -Xcompiler.

Rework loadMachineFile function with usage of the cmake-function site_name(): https://cmake.org/cmake/help/latest/command/site_name.html

As all warnings from VirtualFluidsCPU might be removed, we added a new CI job wich only builds VirtualFluidsCPU and treats all warnings as errors.

Add the option for call linkBoost() without components. In that case cmake will just add the boost header dependency.

Restructure the cmake cuda process. Remove the policy and always set CMAKE_CUDA_ARCHITECTURE to 30, if not set. This will fail when someone is trying to build with cuda 10.2 or higher, as cuda 10.2 minimum support is 52.

NOTICE: preferred way is to set CMAKE_CUDA_ARCHITECTURE as cmake argument OR in the config file to the real gpu architecture. For example phoenix tesla P100 have the compute capability 60.

Change in cmake: When BUILDTYPE is not defined in vf_add_library(), a library is build regarding the CMAKE_BUILDY_TYPE.

no changes in configFiles for numerical tests necessary

Add gcov cmake and a python bindings cmake option (default off). Add the compiler flags also to the test targets.

Changed AppleClang no warning flag from no-reorder-ctor to no-reorder. Disable unroll-all-loops for GNU when IWYU is enabled. Fixed Cycle in include-mapping.

Exchange DEPENDS (which links everything private) variable with PUBLIC_LINK and PRIVATE_LINK. Therefore VirtualFluidsCPU is linking everything public know. Everything else is still linking private.