Build Docker container so people can contribute right away

[rgerkin]

The entire ca1_analysis notebook can be executed without error (except cell 3 because the web api has changed) on any machine with Docker installed with:

docker run -it -p 8888:8888 -v /path/to/your/working_copy:/home/jovyan/work/neurotree jupyter/datascience-notebook bash
export TF_BINARY_URL=https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.11.0rc0-cp35-cp35m-linux_x86_64.whl
pip install $TF_BINARY_URL
pip install fancyimpute
jupyter-notebook

And then the notebooks should be available and ready to run at http://localhost:8888.

The only TODO here is to build a new Docker container that wraps those pip install lines into the Docker build so that the container can be entered and used for running/editing notebooks without requiring any new packages.