This will ensure that in verbose mode, the final state of the linear
system solvers is always printed. Previously this was the case only when
refresh rate was equal to 1.

Moved tnlObject_impl.h back to tnlObject.cpp since function definitions
in header files cause duplicates when building libtnl. The constructor
of tnlObject is __cuda_callable__ so it's empty definition was moved
directly to tnlObject.h to make it visible to nvcc.

Problem with marking tnlObject's virtual destructor as __cuda_callable__
is yet to be solved...

Parentheses around operator& are necessary due to operator priority

Used shorter __cuda_callable__ instead of __host__ __device__ and
testing if HAVE_CUDA is defined.

RowLengths was renamed to CompressedRowsLengths.
OpenMP support wads added to matrix-vector multiplication and to scalar
products.
CSR format can use Cusparse for SpMV.

Adding CUDA architecure and CUDA cores to log.

simple-solver and make-project were removed from examples.

Scripts tnl-compile and tnl-link were added.