Option to load the notebook either in JupyterHub or Binder (as backup). Loading of data should use `urlretrieve` or similar, so that the notebook will run even in the absence of the repo itself.