When Peter Karbe, chief lens designer at Leica, was asked about the resolving power of the S2 lenses he said they didn't test it. He just mentioned (with a confident smile on his face) that from the specs it had to be somewhere in the range of 300 lp/mm in the center of the IC (probably at f/4). I'm sure the Rodenstock and Schneider digital lenses have similar resolution that is virtually diffraction limited in the center at their specified working aperture range.
For the sensor manufacturers that means they could go to down to a 2 micron design which would still improve image resolution in the center of the IC with these lenses for high contrast detail that would be barely resolved as very low contrast image detail. But there wouldn't be much difference in the corners of a large sensor compared to a 5 or 6 micron sensor. For critical corner performance you tend to stop down, probably to f/8 or f/11 where you're in diffraction territory anyway. In that case a smaller sensel pitch would help you to resolve the diffraction pattern a little better. It's always better to have four pixels for an Airy disk than just one. It would also help to overcome the limitations of the Bayer array and retain the color depth at high magnifications. But reducing the sensel pitch inherits other problems that might affect the overall image quality more than the slight gain in resolution. It's all about balance. Increasing the image format would be a smarter move to improve overall IQ with current technologies.