WIP: Adding a JupyterLab instance to the Jupyter software release
The goal of this MR is to add a JupyterLab server process to Jupyter instances, so that users can experiment with the new platform.
The initial commit installs all required jupyterlab eggs via PyPi and creates a new service script that is configured identically to the regular jupyter-notebook one with the exception that it calls jupyterlab instead of jupyter-notebook and it uses port 8889 instead of 8888. @luke suggested some sort of "crazy" method of hosting jupyterlab on port 8888 but on another web address which he should elaborate on.
Right now jupyterlab is configured identically to jupyter-notebook and both the Python2 and ERP5 kernels work fine. Loading ERP5 notebooks can be slow, though, because unlike regular notebook servers, jupyterlab keeps all notebooks in a single browser tab and has its own tabbing system. And it seems that jupyterlab's Javascript waits for the entire notebook to be loaded before letting the user do anything. In other words, it freezes until the entire notebook is loaded. This seems to be an issue of Jupyterlab so hopefully one day it will be fixed.
Everything should be completely usable in this state so I'd really like for someone to test this out a bit more.