Nobody knows better all the technical details than Matthias.
I was thinking on the, apparently at first sight, "simple" solution of adding a single SPDIF / AES (or ADAT) input / output which would allow the connection of an external "cheap" converter (like an ADI-2). But I'm sure this has been already evaluated by the minds behind RME. This would not only allow monitoring but also talkback for a relative low cost for us the end users.
But even this solution might be tricky, as clock sync and other functions would have to be implemented in the FPGA (inside the ExpressCard (1)), and RME might be already resource limited (and even space or heat limited due to the ExpressCard specs)! There are a lot of channels to freely route and mix there! Maybe next year, when Xilinx releases an improved Virtex we see a rev.2
The other option would be for RME to release a new ADI-2 with MADI support . That would be NICE. As Mr. Flock said in the interview, in the near future MADI will be everywhere. Let's hope so.
(1) As MC suggested on the other thread, it seems that MADI data is just passed thru the firewire cable to the madiface express card where the "real" processing takes place.