As much as I encourage extension, that piece of logic is not final for no reason. If you were to change the logic used there in the way you propose, you would effectively create a conflict with any subprocess nodes that get added to the diagram. The node you are adding *isn't* a subprocess (it's a customized *ServiceTask*) and you shouldn't therefore generate ids that end up in the same ID "space" as real subprocesses.
I don't see a reason to prevent people from doing this, so we can open it up (I propose a way to provide your own ID generator classes, similar to how it's done in JPA), but you should realize this has potential problems if you use it and don't make sure you won't get conflicts.