> a very strong assumption is that gas ratios do not change through processes like diffusion

It seems to me this should be an easy matter to analyze in detail. Diffusion would blur two distinct bands into each other, right? A sharp change in gas composition would appear more gradual with time, as diffusion blurred the boundary. So, are there sharp changes in the old ice? Lack of sharp changes might not prove anything, but the presence of sharp changes in very old ice should be informative. I'm no statistician but it seems to me a statistician should be able to look at the data and give you an upper bound on diffusion.