Refactor and image optimization

Changes and refactor for v0.3.3

Description

• Added hub https://github.com/github/hub
• Create persistent environments
  • conda create -y -p $LOCAL_ENVS/tabsAreGreat python=2.7 ipykernel 'scipy==0.16.0'
• Discover persistent environments
  • conda env list
• Create a new kernel
  • $LOCAL_ENVS/tabsAreGreat/bin/python -m ipykernel install --prefix $JUPYTER_WORK --name tabsAreGreat --display-name "tabs are great"
  • Refresh main page
• Added ML modules
  • Tensorflow 1.6
  • Theano
  • Keras
• Optimized image size by starting from ubuntu base
• Build scripts now work on OSX
• Added project and partner spaces

Upgrade to Jupyter v5.2.0

Changes

• Added PYTHONUSERBASE variable, which points at tacc-work/jupyter_packages, and allows for persistent local pip installs
• Added pyemd
• Added graphviz, python-graphviz, cmake to support DSGRN
• FIXED git funtionality by setting
  • GIT_COMMITTER\_NAME=$JUPYTERHUB\_USER
  • GIT_COMMITTER\_EMAIL=${JUPYTERHUB\_USER}@sd2e.org
• REMOVED the hardcoding of MPLBACKEND to Agg
• Refactored images to start from taccsciapps/jupyteruser-base
• Removed jovyan user and environment
• Optimized docker files for size (now smaller)
• Updated python stack
• Numpy now uses Intel MKL
• pip2 now works
• Updated probcomp stack and examples
• Updated sbcl to 1.4.3
• Removed Spark, pyspark, and example
• Updated instructions for contributing and maintaining software
• Updated Makefile to utilize build/build_jupyteruser.sh to:
  • Check for dependencies
  • Check tags on docker hub
  • Version each sub-image
  • Build, test, and deploy without squashing tags