Investigate why some visualizations cannot be captured/restored

[thinkh]

It seems that some examples, e.g. Reorderable Matrix, the visualization state cannot be captured and restored.

Hypothesis
Vega supports Group Marks, to support a Multi-Coordinated-View-like setup.

Group marks provide their own coordinate space and can include nested data, signal, scale, axis, legend, title and mark definitions.

The function view.getState() captures only signales and data attributes that are available at root level (and not at group level).

Hence, we need to define how to push the signal/data to the root level so that it can be captured and restored. This part must be described in the paper as well.

[mstreit]

It makes it more complicated but it's an interesting aspect how we can handle MCV in VEGA.

[thinkh]

Does not have a priority at the moment, since we decided to use an adapted Gapminder as example in the paper.