update packages to allow for interactive mpl

[bolliger32]

Workflow

• [x] Closes issue #77
• [x] Notebook:worker pairing builds & passes local build, visual inspection & client.map tests
• [x] Passes travis tests
• [x] Image deployed on dev branch (see notebook and worker dev tags)
• [ ] Worker passes manual deployment test on compute.rhg
• [x] Notebook+worker passes test-hub deployment
• [ ] Updates integrated into downstream images

Summary

Updated conda packages with the original intent to allow the %matplotlib widget magic to work, enabling interactive mpl plots. In doing so, ended up updating a few other packages (some of which were necessary to enable this). See list below.

Also, conda installed latest pip, rather than conda installing earlier pip and then pip-installing latest pip. If that was there for a reason, just let me know and I can undo that change.

Features

• install ipympl
• update:
  • jupyterlab
  • matplotlib
  • numba
  • numpy
  • jupyterlab-manager
• pin some labextensions

[bolliger32]

ok @delgadom - bumped the rhg_compute_tools version on worker and moved those pip installs back to the bottom of the docker file

[delgadom]

This and #77 look good to me. these ready to be merged?

[bolliger32]

I think so. Did we ever figure out why numpy was 15.1 instead of 15.4?

[bolliger32]

conda list says 1.15.4 on the compute-test notebook, so maybe something is going on with the latest build of numpy

[delgadom]

@bolliger32 I actually can't reproduce the numpy version issue. just spun up compute-test.rhg.com and found numpy 1.15.4.

[delgadom]

and I checked the docker build log and that's what it says it installed. you saw this on the notebook server? do you have it installed with pip install --user somehow?