The previous Beta version had some experimental kite logic. However, it did more harm than good for most people due to the nature of trying to run generic kiting logic.
The issue comes down to needing to design CR logic around various tilesets, rather than trying to make code that works across them all. In large open tilesets, you can easily kite around mobs and the logic is pretty simple. In small closed tilesets, you can't. If you have a build that needs kiting, you need really specific logic to kite in the small closed areas that would otherwise interfere with everyone else.
There's two ways you can implement kiting logic. You can either try to keep moving "backwards", away from the mob until you back yourself into a corner, and then try to re-position, or you can simply move back to a fixed locationn that was deemed safe. AreaStateCache.Current.CurrentAnchorPoint exists for that reason; it's the point you first spawn in an instance, and in the old logic, was used to move back to when you needed to kite.
In both cases though, there are various combat scenarios you need to code around. Around doorways, regardless if there is/was a door there or not, if you kite out of LoS, your default logic would then try to get into LoS by running back into the room. Your kite logic would then trigger again, if you didn't limit it, and you'd run back and forth trying to do a hit and run, but mostly be unsuccessful. This happens a lot in certain tilesets and really diminishes the purpose behind kiting.
So for now, the ExampleRoutine provided doesn't try to do any of that stuff, but it's not meant to be a final solution for CRs. The combat model is tricky in this game, and trying to write one routine that handles everything correctly is quite an undertaking. From a design perspective, a much more advanced CR design is needed to be able to handle this issue and others correctly in context of the game.