Better separation of Model from View.

[TomasMikula]

StyledTextArea is a Node and as such takes care of rendering the visual representation of a document. In other words, StyledTextArea is the view. A big part of the model is represented by EditableStyledDocument, which is independent of the view. However, some parts of the model currently leak into StyledTextArea; namely undo/redo, caret position and selection. As a result, for example, to test undo/redo, one needs to instantiate the view, even though undo/redo itself doesn't depend on the view.

I propose to introduce a class (StyledTextAreaModel) to encapsulate the model.

[JordanMartinez]

So the proposed StyledTextAreaModel would have an EditableStyledDocument and it would move ESD's plain and rich text changes EventStreams to the model itself. Then StyledTextArea would be passed the model itself as Constructor param (and possibly some other things), right?

However, due to #152, and in providing the foundation for #222 to be implemented, if caret position and selection are stored in the Model, how would each View know what to display? Would the Model have a list that stores the caret positions and selection IndexRange of each StyledTextArea that displays the model, each of which is bound to its corresponding StyledTextArea property?

[TomasMikula]

EditableStyledDocument would stay without change. StyledTextArea's plain and rich text changes would move into StyledTextAreaModel and StyledTextArea would just delegate the corresponding methods to model. Model would not be shared, each area would have its own. However, EditableStyledDocument could be shared by multiple models, as it is now.

[JordanMartinez]

Ah, ok that clarifies a lot.

[JordanMartinez]

See my PR (#242)