Thanks for the advice about the dev board. My plan is to begin experimenting with something slow and simple, so that I can work out what specification we actually need - Number of LUTs etc., then when I get that ready, I can acquire more suitable hardware.
Sam3u looks interesting, but quite a lot more complex than Fx2's. Why does it appeal to you? It seems to me that I can expose the tail of the compression block as a parallel bus, and pipe that through the Fx2's GPIF into packets. Rather simple. What do you think?