Adding support for Convolutions in CUDNN backend and GenVISC

llvm/include/llvm/IR/IntrinsicsVISC.td

llvm/lib/Transforms/DFG2LLVM_CUDNN/DFG2LLVM_CUDNN.cpp

llvm/lib/Transforms/GenVISC/GenVISC.cpp

llvm/test/VISC/DNN_Benchmarks/common/include/tensorUtils.h

llvm/test/VISC/DNN_Benchmarks/common/include/visc.h