Main enhancements:
* Cleanup and alignment of Table and Array editor code
* New option `array_copy_in_place` determines whether a copied element is
  placed after the element it is copied from, or placed at the end of the array.
  For `array` the default value is `false` corresponding to the existing behaviour.
  For `table` the default value is `true` corresponding to the existing behaviour.

Rather than the potentially expensive validation of a value
against multiple schemata, how about we provide the path to
a value field that tells you the type and an array of values in
the order of declaration!

This change removes the data-index item, and instead discovers the index of the currently active tab - this allows
the tabs to be moved without refreshing the value - making
move, copy-in-place and drop WAY faster (not here but in
a subclass)

Also a fix to the hard-usage of the index in the editor path.
As before, instead of using the row index we use the 'editorId' set by the class/subclass. (in array.js that IS the
index, but in a subclass that, say, forgoes the use of the cache and, instead, uses editor paths NOT tied to the array
item order, the implementation of move, delete, copy, add, and drop can all be vastly improved, and in a way that does not degrade with the increasing size of the array).

The tiny increase in resources is justified in order to allow
subclasses to use a different strategy for row order management
(See FastModArrayEditor)

All button event listeners now call an instance method for the
main implementation, allowing subclasses to implement different
strategies (e.g. see FastModArray). Those event listener MAY
choose to return a value that indicates to the event listener that
the remaining functionality of the listener should be executed.

Mostly that's specifically returning `true` - which is backward
compatible to the original array semantics. For one case, where an
object is expected, _not_ rturning that object has the same effect.

(and this doesn't fix it)

…led for

If run standalone - there appear to be a lot of
timing issues if run in the full suite

It's not worth the trouble they cause!

The test case no longer uses unnecessary popups

(In particular, being able to remove button labels in subeditors)