Fix startup script to use python from the image

This is a bug fix release to modify to make sure the Python script that configures the environment of the user is always executed with the python (2) that is installed in the image.

Bug fix: ignore PYTHONHOME when defining kernel environment

PYTHONHOME was recently added to the environment of Python3 LCG releases.

If we do not ignore that variable when generating the kernel environment, a python2 installation will look for modules in the CVMFS python3 installation.

Quoting Diogo:
"
This python is always Python 2. But... it changes according to the stack: if the stack if Py2, it uses the LCG python2. It the stack is Python 3, it uses the system python2.

The -E ignores the python variables as you say. Otherwise, when using Py3 stack, it was looking for python2 modules in a python3 path.
"

Octave kernel + truststore for NXCALS

This release:

• Enables the Octave kernel.
• Adds more control over the kernels that are shown, so that only the ones that are tested will appear in the list. The addition of a (non-tested) new kernel in CVMFS will not pollute the list of kernels.
• Creates a truststore from CERN certificates that were already installed in the image. This is required by the NXCALS to connect to a server with a REST API, where they obtain some metadata they need before accessing Hadoop/HBase.

Backport of the Octave kernel

This release enables the Octave kernel for the 2.x branch, which is in production at the time of releasing.

Selection of kernels to install from CVMFS

This release fixes the issue of having kernels (Octave, Go), installed on CVMFS, that appear also in SWAN without explicitly copying them.

New interface and Spark integration

Release that supports the new SWAN user interface and the integration with Spark clusters.

Detailed changes:

• Replaces Jupyter FileManager with a SWAN customized one (to enable "Projects").
• Points default URL to the new home endpoint (projects).
• Adds configurations for Spark integration, as well as the nbextensions SparkMonitor and SparkConnector.
• Adds beautiful soup python module (for SparkMonitor).

Remove Spark notebook cloning into CERNBox

Minor update release to avoid cloning a sample Spark notebook into the users CERNBox.

Extra save and custom notebook extensions

This release incorporates the following changes:

• New extra save for Jupyter notebooks.
  • This keeps a local copy of notebooks to be able to recover them in case of session failure.
• Nbextensions deployment.
  • This allows to deploy nbextensions from the host machine into the container.