If you start with the old film format diagonal = the "normal" focal length, I think that you can go as short as half that distance with a flat film plane camera design with reasonable expectations. The rest is just math.
I'd say that sounds about right. Some of these formulas that use a single magic constant have just combined a suitable light wavelength and other parameters to keep things simple. As Jim notes, and I have commented on in the past, the Mr. Pinhole site uses some assumptions and isn't very helpful as to telling the user what they've assumed. I have found the Mr Pinhole coverage numbers to be somewhat pessimistic relative to my experience. (I usually use Pinhole Designer, but that is only available for Windoze.)
It is inevitable there will be fall-off away from the central axis, the pinhole is becoming an ellipse viewed from off to the side (and closes altogether viewed from 90º off). But an additional cause of fall-off is having a thick pinhole plate. One wants a very thin plate and a clean knife edge to the pinhole.
The main thing to remember is that you can be considerably off from these "optimum" numbers and still get results, maybe even results that you like. The article Ralph posted mentions a slightly different optimum for contrast versus sharpness; so even what constitutes "optimum" is a bit rubbery!