Okay. I've done some testing. Here are my results.
Running UVCONFIG.EXE manually solves the problem of it claiming to not have enough conventional memory, but will usually detect the wrong video card.
Using a newer version of UVCONFIG.EXE to force-select an nVidia RIVA-128 (the last generation of nVidia cards to use VESA for any high-resolution modes -- eg, before Windows 95 was popular) does not work. When snapping in the UNIVBE.DRV created by the newest (free) version of SciTech Display Doctor, 6.53, Constructor fails saying that it is an 'old' version. When NOLFB is run, UVCONFIG detects a Winbond video card. Without it, it detects an InteGraphics card.
SciTech Display Doctor 5.3a [DOS only version] complains about being unable to open an input file after it has extracted its files.
The game also complains about not being able to find its CD (that is, once you get past the SDD problems). Using SAPUCDEX fixes this, although I'm uncertain why it occurs in the first place since the game just accesses the hard drive on a full install (which is what I have). Once you actually get into the game, at least on my video card, the graphics were all stuck in little green bars in the top of my screen. This is not unsurprising, but what WAS surprising was that after the 'high-resolution' logo, the intro video ran looking correctly. (This is probably due to the fact that it is low-resolution.) Of course, once it went back into high-res VESA mode, the video was stuck in green bars at the top again and I had to force-quit out of the VDM.
The Windows version is equally plagued with problems. It displays a few white pixels, and after hitting keys, it displays some other coloured pixels on the same line, but never produces any other display. It eventually crashes back to the desktop (the background seems to be in 256 colours after the crash).
I'll keep working on it and see if I can come up with anything. At this point, however, it isn't looking very hopeful.