Actually their wording jives with the way we've done it pretty much all along - multiple attack powers don't trigger marks as long as at least one of the attacks goes against the marker. I know that's not canon but it works better for us. For instance, it encourages solos to spread the love instead of delivering all of their attacks (or none of them) at the defender.
Concerning Divine Challenge in Particular:
I'm not particularly in favor of that change of various reasons. First, and most critically, it's inconsistent with the basic marking mechanic. Now, you've changed that
too, but WotC hasn't: the results is Not Good. Secondly, most monster multi-attacks consist of several other attack
powers so even if a power is the granularity at which a mark works, it's too complicated to decide
which power. Read straightforwardly, in fact, a power which happens to be used as part of another power is still a power - so such (very common) multiattack powers would still trigger on each attack.
All in all, that's very poor wording. Either
marks should be changed to not trigger unless the "outermost" power fails to include the defender (and then with
examples to illustrate what they mean), or the challenge should stick to the normal marks.
Concerning marks in General:
As a matter of design, it's preferable not to make rulings like this depend on purely-metagame packaging issues. Sure, I could live with the change you're proposing, but it means that two otherwise identical attacks may play out very differently depending on whether they're packaged as one power or two - that's not very intuitive. Also, by encompassing multiattacks you're exposing the game to unnecessary complexity when the attack vs. the defender is not the first attack and may be prevented by an interrupt. It turns positively quixotic if the defender does the interrupting: can an assault swordmage teleport in the way of an enemy when the marked enemy does attack-move-attack but attacks an ally first? What if the enemy intends to attack the swordmage but decides not to (walks around the corner, sees juicier target...)? Do you "roll back" to the original attack? Because the -2 penalty applies depending on a factor that may be
later in time, you've got a temporal paradox.
E.g. take the assault swordmage and a marked enemy using an attack-move-attack power. The enemy intends to attack two allies and thus the swordmage intervenes after the first attack. But when he does, he teleports adjacent as an immediate reaction: now the monster is adjacent to his marker, and decides to complete the power by attacking the SM. That's a paradox: the swordmage could not have intervened because the attack included him, but if he doesn't intervene, then... he can and will intervene?
Almost all marks already have a limiting factor that makes them "less good" against solos: they can only trigger once (and the -2 just isn't such a big deal; the defender is likely to be one of the least attractive targets despite that). Furthermore, solos are a very poor example since they should be very rare and chock-full of special abilities to make precisely these kind of situations less likely or less relevant. Solo's aren't a good example of general gameplay.
Just say no to temporal paradoxes in rule resolution: keep things working based on instantaneous events (bursts, blasts, or singular attacks), and don't depend on meta-game notions where possible (such as easily possible here).