(And, of course, this makes sense - if you're tracking ammunition, rations, and healing potions, you actually already have a bunch of per-adventure resources anyway. Eventually, you're going to have to retreat to restock.)
...
In the current model, if you structure an adventure with a time limit, and then the PCs have a really bad first encounter (whether because you misjudged the difficulty, they rolled really poorly, or whatever), then suddenly the whole adventure is in real trouble. However, you can at least build in time for, say, 2 ERests, and all is well.
From the dramatic perspective, what I want is that the closer the party gets to the end, the harder it is to back out (and the less they want to). Getting mauled earlier in the adventure, backing out, trying again--can even add to the drama. Backing out right before the boss fight is a much bigger jolt.
Old dungeons got some of this aspect from that operational resource play. It took risks and resources to get from point A to point E. If you were already at D, and E looked questionable, you still might try it. Even if you didn't, you had a reason to at least consider it. When you didn't, it was usually because the threat of E was overwhelming. So the back out to rest was grudging. Certainly, if a partial back out to C was a decent option compared to all the way back to civilization, the party would jump on it. (Go back to the orcs lair, restock on oil and arrows, etc.)
So I guess my one objection to 1 AP per encounter, 1 daily per milestone, or other such solutions is that they don't go quite far enough to address the above. That is why I would much rather tie the resources to milestones and then change how they work. That way, I can make milestones more dense towards the end of the adventure, and make the tougher stuff more rewarding to push through. But if I have an adventure that follows a different pattern, I can move the milestones around to reflect that.
I played in a one-shot at a con once (2E AD&D), where the DM had the worst fights as the opening and closing fights. The opening fight was, in fact, slightly worse than the last, because we weren't expecting it to be that way. Talking to him afterwards, he indicated that this was something he had started doing in cons to address dramatic pacing for people who had little attachment to their characters.
People expected the gradual build up, and thus weren't terribly motivated in the early going--except to save resources for later. Whereas, the way we played his adventure, we were shook out of this mindset early. Then, thinking we were in over our heads, we had to really focus in the middle. Finally, the boss fight didn't have to be that mechanically tough, because we were already drained of resources. So it became about the personalites--and we really cared, because the boss' troops had been abusing us for 3 hours.
I would tie none of this to surges, because besides the mechanical drawbacks, surges are meant to, and intuitively to the players will, correspond to that increasing fatigue. They are the final limit, and are thus a resource to be hoarded in operational play. Really, rather than tie surges to this other stuff, I'd rather move something over into that category with surges. I might feel differently if someone managed to include surges in a complete "reward cycle" of D&D play--but as written now, they are pure resource.