Still the same, performing those works is primarily a feat of brain virtuosity.
I would have to find an impressive display of mechanique in his recordings and compare it with a score to tell just how impressive it is from a mechanical-virtuoso perspective.
See, these are the 2 primary factors people use when considering the difficulty of a piece - the brain-virtuosity difficulty of the sheer complexity of note figuration and disputation, and the 'randomness' involved.
Choosing the 'most difficult' in this area is a rather silly thing to do, as I can easily make a piece with just a few more notes in it and say it's more difficult.
The other, as I've said, is the mech-virtuoso factor, which again isn't completely without problems, to determine a level of difficulty one must predefine a specific tempo.
You could say that there are pieces which are very difficult in both perspectives, but it is apparent that notes and figurations have to be INGRAINED in a pianist's reflexes and muscle memory to be performed at peak conditions.
In overly complex works, pianists simply don't have the time to ingrain every figuration to the utmost efficiency, and they rely on sheer brain virtuosity - which is basically SIGHT-READING from memory, or in some cases - literally sight-reading.
This is why, to determine a pianist's technical ability, we must hear them in pieces which are ingrained in their pianistic subconcious, and in their fingers.
Sure, theoretically this could be done is Sorabji's and Finnissy's longer and more complex works, with ALOT of work, but the music is of such a nature that it would render it pointless.
This is why more streamlined and relatively simpler music is much better suited for a display of technical ability - particularly the Chopin etudes, which exercise each of their chosen technical figurations at length and provide merciless evidence of the pianist's technical abilities.