Ooooo, lots of food for thought.
Ok, if you need XP and have certain versions of Windows 7, then you can download and use XP for for free - http://windows.microsoft.com/en-GB/windows7/install-and-use-windows-xp-mode-in-windows-7. Support will end soon in line with XP, but what the heck, you don't have to maintain an old machine.
I've found that FF runs like a whippet with a Zimmer frame on Windows at the moment - or at least with FireBug. And if I run the same combination on a virtual Ubuntu 12.10 on the same machine its really fast. I'm not quite sure what is going on.
With FlexBox, I think that utilising new technologies in the early stages of their development helps to understand them. It provides knowledge even if older browsers are not supported. Through demonstration of new technology you can provide the impetus for the non-technical managers to make the technical decisions about upgrading browser versions (deliberate Catch-22 here). The decision about upgrading is always a problematic one. You don't want to hang on to the constraints of the past and at the same time plough too far into the future with the increased risk of failure.
I take the view with design that you should de-risk a new technology on common stable hardware and software at the time. Design, implement and simplify the code. Then tackle how you will add fall-backs if required. Often as development takes time, by the time you've completed the finished product then things have moved on anyway. That upgrades have happened, that the light at the end of the tunnel was a new dawn and not an oncoming train.
Themes need to be like advertisements. The new product has to be wanted and more importantly be better than the product it replaces. Use browser detection to provide fall-back looks that make people want the new look. And that look happens by a simple browser upgrade. Problem -> solution. Old -> new. Everybody needs a reason to expend effort, so give them one that is credible.
Cheers,
Gareth