Hello Pascal/Grant,
the change we have introduced did not break any of our existing demos or documented supported cases. I do agree with you - in hindsight we should have announced it somehow. However, the description would be extremely vague and unhelpful; the change we have introduced breaks cases we are not even aware of, nor have ever recommended as an approach. Nevertheless, we do our best to provide the necessary assistance with each case.
I would also like to address some principal problems with the approach in question. Each Kendo UI Widget should be the sole owner, modifier, and event handler of its DOM fragment. Using angular expressions in the markup has worked initially, but will definitely introduce erratic behavior with changes happening after the widget instantiation. The widget can't react to them accordingly (by re-decorating the changed markup, recalculating necessary sizes, etc).
The recommended way to do such changes should be applied through each widget API/configuration options, which we consider quite extensive.
In case you believe that the change has introduced a performance regression in certain cases, please isolate it in a runnable case, using our Dojo service. Obviously, we will need to compare that with the performance before the change, too. Please make sure that you include the profiling results from your investigation.
Finally, I would also like to point out that in many cases wrapping our widgets in custom, project specific directives is the way to go. Doing so provides a clean, simplified directive catalog to work with. The standard directives we ship aim to cover a wide variety of cases, which the developers in your projects may not need.
Petyo
Telerik