I think I know how one might accomplish the same thing as what I am asking, but using an indirect method.
The idea would be to scan the 4x5" film or transparency in three overlapping groups of scans at ~2500 dpi. Each pieces would be scanneded several times with a miniscule random shift between the scans within each of the three groups. One would then use a superresolution program, such as PhotoAcute3 to the scans within a group. The result will be three enhanced resolution (~5000 dpi) scans. These three would then be stitched to make the full high resolution scan (~5000 dpi).
A fly in the ointment is that PhotoAcute3 has not been sold for several years. One can still download the program, but it is an evaluation copy that puts a watermark on the image when it is saved. Another thing is that leafscan is a very slow scanner, so by the time one does all of the scans a long time will have elapsed.
By the way, yesterday I tried PhotoAcute3 on 8 repeated scans made with a canon FS4000us scanner, and there was a very noticeable improvement in resolution of the image. This was most obvious in random lint fibers on the negative, which were much sharper after the superresolution process was applied. This was not a result of conventional sharpening, but comes from the fact that if you scan the same image several times with very small random shifts in the image for each scan it is possible to synthesize a resolution enhanced image... that's true resolution, not just sharpening.