Apple wants to drive away developers who are spreading their apps thin over multiple platforms. I think you're right that it's quite unlikely they will relent on this, at least on paper.
As for policing, I think you're also right that it won't be particularly onerous. I see enforcement less to the letter and more to the intention of 3.3.1: If your submitted app feels like a rough port, they'll probably check for cross-compiler evidence and reject it on those grounds, but if it fits right into the iPhone experience, they probably aren't going to go hunting for traces of Lisp.
What is has to do is with lock in. Right now, there's a decent barrier to porting an application from iPhone to Android, and in most cases the port will cost more than the profit it makes, so it won't be done.
If people start using multi platform engines, the ports will be very cheap, and the iPhone will have far fewer exclusive apps, which is a huge threat to their platform. Each multiplatform app reduces the costs of a user switching away from the iPhone to an Android phone or similar.
Apple's been on the losing end of lock in with Mac OS for a long time. I think now they're experiencing the effect in their favour and want to keep it going for as long as they can.
If everyone is writing in Cocoa on XCode, Apple's latest & greatest will be deployed that much more quickly.
I think this interpretation is consistent with Jobs' terse emails here, particularly the line, "intermediate layers between the platform and the developer ultimately produces sub-standard apps and hinders the progress of the platform" (emphasis mine).
Suddenly it matters very much indeed that lots of games start calling new platform-native APIs, right fucking now.
Saying that they're doing it to further their platform sounds a hell of a lot better than saying that they're doing it to hurt their competition!
At best you could say that there won't be an intermediary platform delaying the process...but I'm not sure if that really is a compelling reason to justify this decision.
Without cross-compilers, the lag is as you say: developers update their apps. With cross-compilers, the lag balloons: the makers of the cross-compiler toolset need to incorporate the latest Apple features (if they choose to do so given the rest of their market), roll out their tools to developers, then developers have to use the new tools to update their apps. Slow!
I would agree with you that Apple's been doing the same lock-in with Mac OS for a long time, but I'm not so sure they've been on the losing end of anything. I remember that when Jobs came back to Apple, they had free WWDC videos online, and attendance was very much in decline. Since he came back, they removed the Java bridge stuff, Carbon and at the same time they're charging good money for a WWDC that tends to sell out really quickly. Lock-in on the Apple way seems to me to be a fairly successful strategy.
That's true today because they're the main platform. This might change if, say, Android becomes much more widespread than iPhoneOS. When this happens, they might concede on things like 3.3.1.
Then again, Apple is quite comfortable with having minority share but huge profit margins in a market: I'd guess that 90% of laptops sold over $1000 are Apple, and that their margin is something like 5 times Dell's...
If multi platform engines take off, we are a few short years away from a situation where the choice of mobile platforms is like the choice between browsers is now - no lock in, low switching costs - and I think they want to avoid that at all costs.
There is already a cross-platform, 100% uncensored dev framework, with excellent distribution channels, and it's called HTML5. I'm not sure there is a niche for a 2nd one, especially if Apple is unhelpful.
Apple isn't focused on developing a smart phone operating system. They're focused on redefining what a computer IS, and how we interact with them. That's why they've been so careful about copy+paste, and later multi-tasking. They are taking this as a chance for a clean slate approach to computing and don't want to be hampered (or their customers to be hampered) by slow-moving third party compatibility layers. How much as Window's development hampered by the fear to break legacy apps compatibility moving forward? Apple isn't afraid to make their developers jump through hoops and make changes if they think it will benefit the platform in the end (a la the Intel switch). They'll provide tools to make that as painless as possible when the time comes (e.x. iPhone app compatibility on the iPad, Rosetta) but having another layer in between only slows this process down.
Whether you view Apple's actions as that momentous and revolutionary is largely irrelevant. I think THEY do.
Not that I'm not upset by this as well, but I'm trying to make sense of it.
As I see it, there are two elements that define the modern software: end-user experience (not only GUI, but things like local storage, geolocation etc), and communication with the Internet. With low bandwidth, Web apps have all the advantage with the latter and are good enough with the former; with ubiquitous broadband, the latter advantage is meling away, and the former just being "good enough" might not be, well, good enough anymore.
Moreover, repeated installations and deinstallations degrade a computer's performances and security. It's much more obvious under Windows than under correctly designed OSes, but always remains true.
I expect a 95% web-based slate/phone to be mostly stateless, and to remain functional indefinitely without requiring administration, as opposed to a computer.
That said, I'd be surprised if Apple spends too much effort policing the rule. The fact that it exists in the first place will probably be enough FUD to prevent third party toolkits from taking off at all. They'll probably do some kind of automated signature testing for Flash, Unity etc and leave it at that. The upside of that is you'd probably be able to sneak a Clojure or whatever app in and nobody would probably care too much.