In that case what you should do is change the output value of the script's resizing function, so that you have it at the desired editing resolution to start with rather than piling more scaling on top of what is already there. You'd have to change the edge cropping and the borders added and so forth along with it, but it shouldn't be difficult to find the right values--if you're doubling the size of the entire image, then they should be doubled along with it, yes? Then you can make another, much simpler script to scale it back down to dvd resolution during the encoding phase.
I don't know if this would affect the size and positioning of the subtitles, though. Actually, if you're rendering a lossless intermediate file, you could probably leave out the subtitles from the initial processing, and add them to your resizing script for the encode, which would bypass any possible issues with that.