Boy, have you got my number! You're very likely right about it not being worth it. I'd say that two out of three gearing projects like this that I've tried in the past haven't worked out for one reason or another. A couple of years back, I spent a fair amount of time tinkering with a three-cog cassette meant to approximate the gearing you'd get with a Sturmey-Archer AW. I ultimately did get it to work, but it was basically an interesting (to me) novelty. I eventually made it back into a singlespeed.
Funnily enough, Sheldon Brown--who, as I noted in my original post, had a two-cog freewheel setup--more or less called me an idiot for the three-speed project. I'd emailed him to ask him something about the project--I forget what now--and he responded that it was a waste of time, and that the amount of weight I'd save by geting rid of four cogs and a front derailleur would be insignificant. Ouch.
Regardless of how this works out in a practical sense, it promises to be sort of entertaining and will give me an excuse to take apart and reassemble some freewheels.
As far as changing gears goes, my situation is that I can handle most of the riding I do in high gear. I would need to switch for one hill I ride over often that climbs just under a thousand feet in the course of two miles, with some very steep sections. I'd also change at the end of every ride to come up the 3/4 mile steep dirt road that leads to my house--a climb that I always have to make when I'm at my most tired. It's not as if I'd be getting off the bike and changing gears at every little rise in the road. If I get tired of messing with it, it will be easy enough to make it into a singlespeed--which, as you correctly point out, is likely to be more practical anyway.