I should maybe note where I got lost on this line of thinking:
When I started box building I got most of my information from a couple of books. One was the loudspeaker design cookbook, the other was a really informative paperback from radioshack that's out of print. I've been trying to find it for years because it really does explain the equations used to determine the values a given box will produce, as well as how to do a crude near-field test on a woofer.

both of these books explained the "electrical" efficiency gained when impedence is halved. This is an absolutely true fact, nobody can argue a 4 ohm curcuit is more "electrically" efficient than a 2 ohm curcuit. The problem is that we're talking apples and oranges. The only thing incorrect here is:
"Halving the impedance of a speaker by adding another driver in parallel will effectively raise the sensitivity of the speaker by 3db."
adding another speaker alters the curcuit, not the other speakers on that curcuit. A 4 ohm speaker is still a 4 ohm speaker when you wire it's twin in parellel with it, it's just the implied curcuit that changes.

apples and oranges: the question at the start of this post was simple "1 or 2?" I think we can assume any option in spl would run your amplifier at it's highest output wattage, so we have to try to stick to a given impedence. let's assume 2 ohms. 2 4ohm subs in parellel, 1 2ohm sub, ect. I've specifically built and tested 2 vs 4 and 1 vs 2 in a few cars years ago, with varying results. The only given was with a sealed encloser the number of drivers didn't matter as long as the impedence stayed the same on the amp, and the power was fairly well below the driver's power compression area.

In vented enclosures I yielded a bit better output around the fb area, but it was very sloped and could have been the result of anything from an out-of-spec woofer to the exact location of the mic, ect.