jupyterlab-extension

There is already a notebook:open-with-voila command to open the voila dashboard in a new browser tab. It's in the palette and the view menu, but it would be great to have it in the preview toolbar too:

We could for example use the jp-LauncherIcon:

![image](https://user-images.g

According to setup.py: https://github.com/jupyter-widgets/ipywidgets/blob/4372d90a71934b4954b626aed3a49ca23f3e210e/setup.py#L116-L118

    # Requiring nbformat to specify bugfix version which is not required by
    # notebook.
    'nbformat>=4.2.0',

nbformat 4.2.0 was released about 5 years ago, so I think we can remove this dependency now by deleting these lines in setup.py.

Problem

Currently the only packages containing front-end unit tests are pipeline-editor and services.
It would be helpful to expand the unit test coverage across all other packages in Elyra, as well as add more tests to existing extensions.

This is an umbrella issue - subtasks can be delivered in separate PRs.

Pipeline Editor

• Test Pipeline Service (runtime types, runtimes, runtime images

When creating a Choropleth object with geo_data specified, but without the choro_data attribute, the _get_data method will raise an exception:

leaflet.py:1361, in Choropleth._get_data(self)
   1358     return {}
   1360 if self.value_min is None:
-> 1361     self.value_min = min(self.choro_data.items(), key=lambda x: x[1])[1]
   1362 if self.value_max is None:
   1363     self.val

Elevator Pitch

Right now Jupyter LSP show it's suggestions on every period. I'd like to have a feature that only generates the suggestions when a hotkey is pressed. (Shift Tab, etc)

Motivation

One of my users asked for the feature.

Design Ideas

Is your feature request related to a problem? Please describe.

While you're working on the notebook you can't see which branch you're working on, unless you select the Git extension tab. I noticed that this can be confusing and lead to errors especially for d

When a TF serving model returns a failure instead of a prediction, pass the failure string to the front-end for display instead of the generic http error we see now.