OT: context.forceNotSticky may not be 100% accurate

[scofalik]

(Follow-up to ckeditor/ckeditor5-engine#977)

To speed up work on ckeditor/ckeditor5-engine#977 I've implemented context.forceNotSticky using some naive assumptions that may fail during selective undo.

Reminder: context.forceNotSticky disables MoveOperation#isSticky property during OT if it is set to true. This is needed for undo to work correctly, however, of course, the flag needs to be respected for proper transformation of MergeDelta, SplitDelta and UnwrapDelta.

Right now, stickiness is disabled (context.forceNotSticky == true) when the delta that we transform by has been undone or is a delta that undoes another delta:

if ( history.isUndoingDelta( originalB ) || history.isUndoneDelta( originalB ) ) {
	context.forceNotSticky = true;
}

This is a bit naive assumption that I though of when I considered in what cases the stickiness fail. The better way to solve it might be like with context.insertBefore. This means that we enable/disable stickiness basing on how the transformed delta was affected by the deltas it was previously transformed by.

BTW.
At one point I wanted to remove isSticky and move those transformation nuances to deltas special cases but this would require a lot lot more work than implementing context.forceNotSticky to fix #isSticky support.

[Reinmar]

Such property does not exist any more.