The original LTHR test was different. I assume they changed to this new approach because it yielded similar results and was easier to apply the test.
In the old test, which I prefer, you ran for a couple of miles to warm up. Then, without stopping, you started your watch and started running at your one hour race pace. At the 10 minute mark, you took a split and then stopped at the 30 minute mark.
Your LTHR was your average HR from the 10 to the 30 minute finish mark. When done properly, it can be extremely accurate to within a couple of beats.
But there are three problems with this test. It takes longer. If the runner goes too slow, the last 20 minutes will be lower than actual LTHR. If the runner goes out too fast, HR will creep up the last several minutes of the run and the average will be higher than actual LTHR.
The latter problem is easily fixable. Take another split at the 20 minute mark. The split at 20 minutes and 30 minutes should be very similar... within a couple of beats. However, if the 30 minute split is a lot higher -- maybe 5 BPM -- then the runner was going anaerobic over the last 10 minutes by running too fast and the average would be higher than actual LTHR.
My guess is that through trial and error, Friel discovered that too many runners (and cyclists, remember, Friel is primarily training triathletes) had no ideal of their one hour race pace and were going too fast or too slow. Then, he realized the all-out 30 minute run yielded similar results. It's a LOT easier to tell cyclists, "Go run as hard as you can for 30 minutes."