cudax MVP Example

[jrhemstad]

I'd like a minimal example of cudax:: that illustrates the most important features:

• Allocate and initialize device memory
• Copy data to<->from host
• Manage devices and create resources like streams and events
• Launch a kernel on a particular device/stream

I think it would be appropriate for our example to showcase a new and improved vectorAdd that showcases the following pieces of new machinery:

• cuda::mr data structures for allocation
• cudax device, stream, event data structures
• cudax::launch to launch a kernel
• (anything else important I'm missing)

Let's put this under the cudax/examples directory. In time, I think it will be a good idea for us to mirror some of the examples from the canonical cuda-samples.