Separating python and r kernels into their own conda envs

[SandyRogers]

This PR:

• creates 2 conda environments (one for Python, one for R), and installs them as Jupyter Kernels for the Notebooks.
  • this helps with clashing dependencies
  • the kernels are managed by nb_conda_kernels
• adds a stub notebook for R so that it is easier to contribute a new R-based notebook
• adds static rendering of the R notebooks so that they appear on docs.mgnify.org
• improves the management of the mgnifyr cache file

TODO:

• the test for using jupyter lab in shinyproxy is intermittently failing when running in headless mode.