We need to do some very deep analysis of layout performance bottlenecks and file appropriate issues. I expect we will find 3 main issues:
    - Use of List in interpreters adds overhead, as well as all the various individual method calls
    - Nodes are claiming they need to be invalidated in terms of layout when really they haven't changed
    - Recomputing preferred / min / max size information when it hasn't changed

I also expect this to be a problem particularly with Text / Label.

However as these are just guesses based on past experience in this area, it may or may not be applicable. This task is to do the full in depth analysis and really find out where we can improve our layout performance.