I agree with some of the earlier posters that it doesn't make sense to talk about whether 3e is "more simulationist" than 4e or vice versa, if "simulation" is defined as simulating the real world. After all, neither system comes even close to simulating the real world (even the parts that aren't explicitly magical).
So I think the appropriate definition of "simulation" to use is not whether it simulates the *real* world, but whether it simulates *a* self-consistent world. Of course the problem with this definition is that *any* set of self-consistent rules by definition simulates a self-consistent world, if you just assume the rules are what govern the world.
A better way of thinking about it is to imagine: "what would the world that the rules simulate actually be like?" True, you would have absurdities like healing to full after a night of rest, etc. But I don't see that as really a problem, because we already accept that things like that are possible because of magic, and if the whole debate about what parts need to be realistic or not just boiled down to which things have the label "magic" on them, I don't see why it would be such a big deal.
But there is another category of things that are issues, and that is: what do the rules imply about what is a part of the world? That causes other problems. For instance, consider the power "Own the Battlefield" (another power that causes problems similar to Come and Get It) that moves every enemy three squares (I think) but only works on enemies, even if you wanted to do it on allies. The most commonly cited problem is that there's no "explanation" for how it works. This doesn't bother me; there's no "explanation" for how magic works, either (and if this were a wizard instead of warlord power it wouldn't be complained about as much.) The bigger problem is that it implies that "whether you're an enemy" is always clear. For instance, let's say you're in a busy tavern and you think there's an assassin disguised as a regular patron. Activate Own the Battlefield, try to move everyone three squares to the left. The one who moves is the enemy. I can think of lots of situations (more than two sides, a Mexican standoff, all-out confusion) where it's not clear who the enemy is.
3.5e also has examples of this, in particular the ability to detect alignments. That implies that, for example, "evil" is objectively measurable, which supports an ethical theory that not everyone would agree with.
A while ago, I wrote another post along these lines, more focused on combat and powers specifically, here:
http://www.enworld.org/forum/general-rpg-discussion/255663-power-system-combat-rest-game-2.html
(scroll down a bit)
Another reason I think that these "simulation" discussions often end up with people talking past each other is that everyone has very different ideas of what is important to simulate, as well as very different experiences that inform what they notice. For instance, I'm a computer science and math student, and a lot of computer science and math is figuring out what you can deduce from a set of information. So I tend to follow those chains of implication ("if you have Goggles of Aura Sight which measures hit points, and there's a power that does a constant amount of damage equal to your wisdom modifier, you can figure out how wise someone is by having them use that power and measuring the damage output...")
On the other hand, I don't really have much experience with combat or medieval history, so I don't notice things related to that ("tactic X was mainly used to help people using weapon Y overcome their inherent disadvantage against armor type Z, so if Y doesn't have a penalty against Z there's no reason tactic X should even be in the game").