* Adding implementation for copying a vector to tensor
* Changing Tensor test to access gpu memory indirectly