Tidy-up CUDA implementation of IProductWRTBase and add CUDA kernels with additional parallelism

This MR tidy-up the previous implementation of the CUDA version of the IProductWRTBase operator and introduces new CUDA kernels with additional parallelism across quadrature points.

Operators/AssmbScatr/AssmbScatrCUDA.hpp

 #pragma once
 
+#include <MultiRegions/AssemblyMap/AssemblyMapCG.h>
+#include <MultiRegions/ContField.h>
+
 #include "MemoryRegionCUDA.hpp"
 #include "Operators/AssmbScatr/AssmbScatrCUDAKernels.cuh"
 #include "Operators/OperatorAssmbScatr.hpp"
 #include "Operators/OperatorHelper.cuh"
 
-#include <MultiRegions/AssemblyMap/AssemblyMapCG.h>
-#include <MultiRegions/ContField.h>
-
 using namespace Nektar;
 using namespace Nektar::MultiRegions;
 using namespace Nektar::Operators;