ScatterWidget return labels instead of colors

[Dristro]

Noticed that the .data_as_pandas returns a df containing both colors and labels. Whereas the .data_as_X_y returns colors, which isn't very useful when interpreting and working with the data.

Update: Modified the ScatterWidget class so that .data_as_X_y now returns labels instead of colors.
Question: Was there a specific reason for returning color instead of label? Happy to discuss if there's a particular use case for it!

[koaning]

I need to think about this one a bit. I don't mind also having the labels in there but I think we want the colors too if we want to make matplotlib charts with the same data that have the same colors as the drawing widget.

[Dristro]

I see your point. Would it help to introduce two separate properties .data_as_X_y_colors and .data_as_X_y_labels? It’s a bit redundant but would allow for both functionalities.

Alternatively, .data_as_X_y could be a function with a parameter, something like type='colors' or 'labels'. That way, the user can choose what they need dynamically.

From my experience, I tend to use labels more than colors, so having that flexibility would be useful. What do you think?

[koaning]

I guess it makes sense to return actual labels for data_as_X_y. If the user wills it, the dataframe variant still can contain the extra column.

Could you add a unit test for this PR?