Document how to setup ec2 instance in AWS to run DeepProfiler

@jccaicedo and I concluded that [docker](https://www.tensorflow.org/install/docker) is the way to go here/

@michaelbornholdt may need to experiment a bit but @jccaicedo feels it should mesh well.

That doesn't solve the problem with the primary issue that @michaelbornholdt was facing, which is installing the NVIDIA driver. The [docs](https://github.com/NVIDIA/nvidia-docker/wiki/Frequently-Asked-Questions#how-do-i-install-the-nvidia-driver) say that the recommended way to do so is to install `cuda-drivers` using a package manager.

But instead, a simpler, faster route is to use a [DL AMI](https://docs.aws.amazon.com/dlami/latest/devguide/what-is-dlami.html).

Note: @jccaicedo had previously maintained [this](https://github.com/broadinstitute/imaging-vms/blob/master/juan-tools/juan-tools.sh) but now recommends the approach above instead.

@michaelbornholdt if all this pans out as planned, please update the [wiki](https://github.com/cytomining/DeepProfiler/wiki/1.-Installing-DeepProfiler) with your notes.