Gas Mileage:
I think the 2.5L and 4.3L are a wash. I'm expecting about 15 to 20 MPG from either one for normal driving.
HP/Torque:
The 4.3L wins here, hands down. I'd never need more power than the 4.3L can put out in stock form (or want it given the strength of my front axle). While this will not be a hardcore wheeler, it will get used off road. I can compensate for the 2.5L's relative lack of get-up-and-go with the planned gearing and recovery gear off road (I actually like winching), so the highway is the main place the 4.3L has the edge.
Off-road Performance:
The 4.3L will be a TBI or MPFI configuration, which again gets the nod. A carb just can't compete on that level, although the simplicity of the carburetor for field repair cannot be discounted. As has been indicated in other threads, a MPFI engine takes a lot of sensors and electronics to repair. I can repair your average carb with bubble gum and bailing wire.
Installation Cost:
On the surface, they seem about the same, but the devil is in the details. Both will be relatively high-mileage engines, although the 2.5L will probably have slightly fewer miles. I would take the stock wiring harness for the engine and modify it for use in the Rover. The engines cost about the same to buy. Engine mounts cost the same, although the 4.3L mounts would require the removal of the stock mounts while the 2.5L mounts bolt on. The 4.3L would require another, larger radiator, while the 2.5L would be able to use the new Series radiator I have sitting in my garage. The 4.3L would probably require replacement of injectors and sensors for reliability, while the 2.5L would probably need the carb rebuilt, plugs, and a cap and rotor. The 4.3L dual exhaust will probably cost more in terms of man hours and supplies as they route it around the narrow frame. It's very likely they can use the stock Rover exhaust as a kind of template for the 2.5L. Overall, I'd have to say that the advantage goes to the Iron Duke here.
Installation Ease:
A Robert Davis 2.5L conversion takes about 6 hours to install, and I'll be using their engine mounts. I expect my 2.5L conversion to take longer, as I am going to replace the entire wiring harness at the same time. A lot of the engineering has already been done, although the custom nature of the transfercase and transmission might add a few hours of messing around. The 2.5L is also physically smaller, which should give me options when it comes time to install accessories under the hood. The 4.3L would require a little work to find the right mounting location for weight, physical size, and cooling considerations. Wiring for the various sensors and indicators would also add a bit of time that wouldn't be as extensive with the 2.5L install. Advantage, 2.5L.
Parts Availability/Sustainability:
We're talking about GM products here, so parts aren't very hard to find in the US. Both engines were used in a variety of platforms, so your average auto parts store at least has them on the books, if not in stock. The 4.3L is a more modern engine, so it will be easier to find parts for and will be supported longer, but that's a slight edge considering the consumable parts are fewer on the 2.5L and many interchange with other engines. We're talking about a rig that will see less than 50,000 miles over the course of 5 years by my current estimates, even considering long trips.
Overall:
While a 4 cylinder "feels" right in a Land Rover and they aren't really designed as "go-fast" machines (braking, suspension, steering...), the 4.3L would make for better highway (mountain pass) and off-road performance. The Iron Duke I had in my CJ could hold highway speeds on the relatively fast and flat Idaho highway between Mountain Home and Boise (until heat or whatever would cause it to buck like a stallion), and my CJ was basically a rolling brick in terms of aerodynamics- much like the Series. An OBDII engine will be much more complex, with the partial offset of having diagnostic capability. It's all pretty much a dead-heat to me right now.