Include `nb_conda_kernels` in base JupyterHub conda environment

[rsignell-usgs]

If nb_conda_kernels is included in the base JupyterHub conda environment, then users can create custom environments and have them appear in the drop-down Kernel list in Jupyter.

[craig-willis]

I wasn't aware of this feature. We'll likely need to extend the default base image to support this, which shouldn't be a problem.

[rsignell-usgs]

@craig-willis, that would be fabulous. Thanks! We had the same issue initially on the pangeo on google, and this fixed it. https://github.com/pangeo-data/pangeo/issues/67#issuecomment-357128933

[craig-willis]

@rsignell-usgs I've created a custom image based on jupyter/scipy-notebook with nb_conda_kernels installed:

https://github.com/nds-org/esiphub/blob/master/dockerfiles/Dockerfile.scipy

Is there an easy test to confirm this is working?

[rsignell-usgs]

I restarted my server and now I can see my custom environment "rps" in the kernel list, so it's working!

Awesome!