Option 5: Continue going the way you are, but with the idea of given back some of the work you have done to the fengGUI community.
(possibly as an extension, I could see a lot of people wanting a 'manager' that would keep track of GUI states and such)
(this is just an idea, I don't really have a good suggestion for you; although I am partial to fengGUI)
What I am intending on doing, if I reach the point of needing complex menu layouts etc, is create my own on the fly editor. Essentially you add/remove widgets on the fly, and edit their location and sizes, and when you're done the editor will spit out code for you to use in your program. I've done it for swing, so creating something similar for FengGUI can't be that hard.
Of course, that isn't top of my priority list right now. You could always beat me to the punch and make something like this yourself though - if you wanted to continue using FengGUI. 
I think part of the problem is that not enough people are using the most recent version and contributing back, to make it a stable, easy to use product. Personally, I picked FengGUI because I heard JMEDesktop had Mac issues - and I needed something that would work equally well on linux, mac, and windows. For prototyping I was using external swing, but it didn't seem integrated enough for my final product. I can see how you could probably make it work on an app that didn't require menus and graphics at the same time though.