Follow-up from "Resolve "Add generalized function for graph plotting""

The following discussions from !11 (merged) should be addressed:

• Allow loading of external data (dantro#145 (closed)): @herdeanu started a discussion: (+10 comments)

  • In #31 (closed), some thoughts on DAG-transformed data came up, but I can't recall what exactly was meant by that.

  The DAG support should already be there because the plot function is marked as use_dag and it works with the data object returned after applying DAG transformations. 🤔 @blsqr do you think that we should nevertheless test the DAG support?

• Implement a mark feature to highlight selected nodes: @herdeanu started a discussion: (+3 comments)

  Another idea for the select feature: Should we provide the option (as default) to mark the central node, e.g. by adding a red circle around the node (also configurable) such that it gets visually clear which node is the central one?

  This could be done in another issue, too, I would say. :)

• Implement the open_edges feature, allowing to show "loose ends" when selecting a subgraph for plotting: @jeremiastraub started a discussion: (+4 comments)

  Do you think that the open_edges option in the select interface really is useful? That option allows showing edges between selected and non-selected nodes (edges with loose ends). On the one hand it would need to be reworked, on the other hand it would also make other things easier if we would drop it...

• Allow customizing the colorbar: @jeremiastraub started a discussion: (+4 comments)

  Another issue that came up for @jhege: It would be nice to be able to customize the colorbar. In Jonathan's case, there are discrete node categories (represented by an integer type node property) that should be color-coded. It would be nice to: (a) Show some kind of discrete colorbar, and (b) customize the labels, thus replacing the integer values by category names. This might be useful feature.

  I don't have much experience with that particular matplotlib stuff. First, I think it would be important to be able to pass further kwargs to fig.colorbar. However, it might be necessary to setup (and customize) a ScalarMappable beforehand?

  @blsqr @herdeanu @jhege Do you have experience with that, or any immediate ideas? Do you think this would be useful? Then, I would try to implement it in this MR or as a follow-up.