I'd like to know the test methodology on that test. If it's anything like the common practice amongst ricers, they brought the stock car in hot off the street and did their baseline dyno, then after spending a few hours installing it, put it right back on the dyno cold, which will always show an improvement.
I have a snorkel on my Discovery, and I don't doubt it restricts flow somewhat at full throttle. It's simply too small of a cross section, and long. However, I don't care so much about a couple horsepower here or there at full throttle, frankly. I only care about efficiency at cruise. And with partial throttle, it's not drawing so much air. More importantly, scanguage is telling me it's running closed loop at stoich, and there's just no room for improvement over that.
As regards air temps, I don't have data on that before/after. I know when moving it's only a few degrees above ambient, but when stopped it can range up to 20-30 degrees above ambient. That is with the snorkel, and the reason is simple, the underhood portions of the intake system heat up and radiate heat into the airstream. You also get false numbers because heat conducts through the system and heats up the sensor element itself, and with no airflow to cool it, it gives a false reading. I've experienced this on other cars.
The Land Rover was never sucking hot underhood air anyway, it was drawing air from inside the fenderwell.
Anyway, for me, the snorkel was mandatory, and it's effects on performance (positive are negative) are near immeasurable.