I think Pieter should always apply the rider "based on the limited test results we have seen". The 1942 campaign was not tested as much as we would have liked
But what is "limited"? In the final ~3 months of testing, I was I believe the tester most involved with testing the 1942 campaign vs. the AI from primarily the Axis perspective, and a number of bugs were discovered. Sure, no full 1942 campaign against a person was tested, but that goes for every campaign because they were all updated shortly before release to the point that they required restarts.
I'm basing it on the results I've had and the results I've seen. You can get a pretty accurate idea of what can happen and what isn't going to happen in the coming turns after even 2 turns, which is also why I said our latest test was pretty much pointless to continue because I could rail reinforcements, including static ones, into the area at a pace that you couldn't match as the Axis.
The significant amount of Soviet active units and generous rail capacity can make the 1942 campaign very difficult for the Axis from the start and the AAR results we've seen, both tester and ones like this one, show that crippling the Soviets through destroying units is pretty unlikely to happen. For example: I'd say Tarhunnas did, for what he wanted to do, a good job with destroying units, but even he's nearly back at his 1942 startline in summer 1943.