CUDA support

[Sha-x2-nk]

Hi.

feature: Adding CUDA support to numCPP, where classes and stuff will be same, but kernels are going to run on CUDA GPUs.

Turns out, I already have a project named numC - numPy similar code, ported to C++, and accelarated on GPU. I used my own implementations for the kernels, but for production code, we can shift to thrust. My code is visible on my github repo, under the name MNIST_NUMC -> modules -> numC. I would request you to allow me work on this and we'll get started right away. Awaiting your response.