This is in accordance with our usual policy against freelists, now that working allocators are everywhere. It should also make memarea.c's coverage higher. I also doubt that this code ever helped performance.
