Fixing scalar product and Lp norm in CUDA.

src/core/cuda/cuda-reduction_impl.h

src/core/cuda/reduction-operations.h

src/core/vectors/tnlVectorOperationsCuda_impl.h

src/matrices/tnlEllpackMatrix_impl.h

src/matrices/tnlMatrix_impl.h

src/matrices/tnlSparseMatrixRow_impl.h

src/mesh/tnlTraverser_Grid2D_impl.h

src/problems/tnlHeatEquationProblem.h

src/problems/tnlHeatEquationProblem_impl.h

src/solvers/linear/krylov/tnlGMRESSolver_impl.h