Sounds like you are going to have a good time learning some interesting stuff, but I think you've got your wires crossed a little bit on a couple of things...
The definition of 'lean' is not necessarily lambda > 1, but simply a lambda that is more lean than the engine can tolerate at that particular condition. All the scare stories about 'I upped the boost and the engine ran lean and melted the piston' don't refer to the engine running at lambda 1.2, probably still < lambda 1 - just too lean for the particular condition, and hence too hot at that point.
At low load conditions, running at lambdas>1 is not an issue and will not melt anything. (Of course, the definition of 'low' is relative to your engine and car)
As the load increases the problem with running on the lean side of lambda 1 is that you are in the less efficient region - i.e. you are not using all the air you pump into the engine, so you need to move across to the 'rich' side, and the (roughly) optimum lambda of 0.85 - but in moving from 1.2 to 0.85, you move through the hottest part of the lambda region, so you need to be careful to do this at a load that is low enough for the temperatures not to become an issue.
Alternatively, you may choose to always stay on the lean side of lambda 1. the problem here is that your operating window is much narrower, i.e. lambda 1 is too hot so you have to go leaner, but the engine starts to misfire around 1.25, 1.3, whereas on the rich side you can go all the way to about 0.6 before you start to misfire, giving you much more headroom.
Another issue with running leaner than 1, linked in with the last point, is that engine response tends to be reduced and fuelling control during transients needs to be much better controlled, again due to the reduced operating window.
Also, as Rob pointed out, the major reason for modern engines operating in the lambda 1 and richer area is the requirement to meet emissions legislation, regarding HC, CO and NOx. This is achieved by using a catalyst, which operates most effectively in a narrow regions around lambda 1, and also by minimising the feedgas concentration in the first place. Therefore at cruise we need to run around lambda 1 to keep the catalyst working and avoid saturation of any of the gasses, and running lean of lambda 1 and even moderate loads significantly increases NOx concentrations, and can quickly saturate the cats leading to breakthrough and a test failure. On the other hand On the other hand, allowing the lambda to wander slighly to the rich side of lambda 1 is less of a problem.
Last point - you are quite right in that its not the fact that you use less fuel when running leaner than lambda 1, as the torque drops off - its the fact that you open the throttle more to get back to the same torque point, and therefore reduce pumping losses. The improved pumping efficiency is slightly greater than the reduced thermal efficiency, so on balance you're up. However, as has been hinted at, the leaner mixture is less willing to burn. One way of helping counter act that may be advancing the ignition timing, but also increasing the compression ratio and improving the charge preparation and mixing will help. Unfortunately your idea of fitting bigger throttle bodies is directionally the wrong way here - larger diameter ports will mean slower moving air so worse charge preparation. Assuming you haven't opened out the inlet ports in the head it probably won't make much difference, so my advice is don't bother with the different throttle bodies for the sake of trying to run leaner than lambda 1.
As an interesting, related footnote, GDI (Gasoline Direct Injection) was re-discovered and developed (initially by Mitsubishi but then others) in order to allow the fuel to be injected as a 'stratified' cloud around the spark plug, so that the localised mixture is actually at a sensible lambda, but the overall cylinder lambda is very lean. At the extreme the throttle can remain completely open and the load controlled by injected fuel quantity - just like dirty oil burning truck engines! When running stratified some very impressive fuel economy improvements can be seen. But this technology suffers the basic problem of NOx emissions (caused by the fact that the cloud inevitably has a boundry region which encounters lean combustion), which requires additional NOx trap systems. Since these are not continuously reducing systems, they require periodic 'regeneration' which is achieved by - you guessed it - running rich for a while! therefore the real world fuel economy improvements of stratified gasoline engines are only of the order of a few percent, while the control strategies, calibration effort and extra hardware required to operate the system make the production cost significantly higher. Basically, its just not worth the effort.......
