I've already done a high-level analysis of this and have an idea of what should and should not be made public. There are a few areas where we will run into problems (that is, Scene Builder depends on things we shouldn't make public), but hopefully that should be resolvable. The primary challenge is producing a minimal API that is useful to users other than Scene Builder (I'm very conscious that we shouldn't let Scene Builder dictate the API we make public, as it is likely an edge case).
1) Creating minimal API (without leaking implementation classes): four weeks.
2) Updating 'the world (i.e. Scene Builder, etc)' to use the new APIs: one week.
Total guestimate (with _no_ padding): five weeks.