* Why: it is then explicitly clear in demos what vendor libs are needed and make its easier to use those demos as a template for integration elsewhere.
* Had been broken by f176979f5bd39e78e0c66c27dced3f7c8b2dc49b with its change to use passed in element to DataExplorer rather than internal generation of element.
Some substantial other related changes:
* Switched DataExplorer view to have element it attaches to passed in rather than being created internally (this was important for FlotGraph to set up correctly).
* Also refactored DataExplorer setup somewhat (e.g. got rid of draw function which wasn't needed -- merging necessary parts with initialize)
* Substantial refactoring of FlotGraph so that we correctly work around flot's delicateness regarding rendering (element must not be hidden etc -- see comments in code)
* Also generally cleaner code
* Also rename element class to data-table-menu for greater specificity
* Allows us to get rid of a nastiness (DataTable view was binding to events outside of itself) -- and another step to making the Explorer completely autonomous and non-dependent on HTML.
* It is clear that if Backend is to handle syncing of Documents we need to only be working with one dataset (or: every Document has to reference a dataset -- in which case Backend and Dataset really are just one object ...).
* This also necessitated a nice refactor whereby DataExplorer subviews running off a common Backbone Collection (DocumentList) re-rendering themselves in response to changes in that Collection.
* Use top menu again
* Refactor main area so that data-table-container becomes data-view-container
* Fix up width of flot graph and editor so they line up reasonably nicely
* In addition fully remove remaining references in models (in getRows and getLength functions) to form of backend (this should have been in previous commit).
