What empirical evidence is this based on? Or is it simply meant to be a reminder to 4e players (and some variants of 3E play) that they are not true roleplayers?
I GM a game in which a "long rest" restores all hit points, and all surges and other daily resources. It doesn't follow from that that the players do not have to engage in non-tactical play. For a start, establishing opportunities to take a long rest is not a tactical consideration; it is, in your language, a strategic one. And the players in my 4e game frequently find their PCs engaging situations in the gameworld, including dangerous physical or combat situations, on less than full hit points, or (which in 4e amounts to functionally the same thing, given that 4e combat relies up the players unlocking their PCs' surges) with less than their full suite of surge-unlocking resouces available.
Nor do the players in my 4e game have their PCs kill everything they meet. It seems to me - but perhaps I'm missing something - that a game in which the only reason the PCs don't engage in violence is because they're worried about being hurt themselves would be a rather shallow game, in which the PCs are self-serving pschopaths. The PCs in my game don't try and kill everything they meet for the same sorts of reasons that I, in my real life, have never tried to kill anyone - eg they have moral outlooks towards, and emotional connections with, other people. And stepping it up from the ingame to the metagame level, the players have their PCs do things other than fight because the players are interested in aspects of the mechanics other than combat, and aspects of the fiction other than fighting.
The idea of "set encounters" is also very playstyle specific. I don't know if it is identical with railroading, but it seems to have some affinities.
As best I have been following it, D&Dnext is actually a huge change already for 4e players - at the heart of 4e's healing mechanics is the requirement for players to unlock their PCs' healing surges during combat. This is a big change from classic D&D (I don't know 3E well enough to compare to it) because it replaces an attrition model of hit point loss with a completely different in-combat dynamic, which has both tactical ("gamist") elements and pacing ("narrativist") elements, and locates the attrition dynamic in healing surges and daily healing abilities. So rather than having hit points suffer attrition over the course of the day, the availability of surges to unlock suffers attrition (through loss of surges themselves, and through loss of surge unlocking abilities like Lay on Hands, potions, etc).
D&Dnext doesn't seem to me to have anything comparable to this.
Of course, none of the above is a rebuttal of your idea about how recovery should work. I'm not sure I agree with your argument, but I don't reject it either - I don't have a good enough feel for the dynamics of D&Dnext. What I'm trying to convey is that I don't think that framing your argument in terms of minor/major departures from 4e play is very helpful to setting it out, at least from the perspective of this 4e GM.
I think this needs more unpacking.
For instance, in Tomb of Horrors or any other fundamentally non-reactive dungeon (White Plume Mountain would be another example, wouldn't it?), there is no practical bar to beginning every ingame day, or indeed every encounter, with full resources: you just declare your rest, your expenditure of healing spells, and your re-rest, and the GM ticks off the requisite number of days. It's true that wandering monsters (none in ToH, but it's unusual in this respect) might upset that dynamic, if they ablate more hit points per hour than you recover spells per hour, but particulalry at mid-to-high levels, and if the GM is following Gygax's advice about modulating wandering monsters to PC activity, that is a fairly big "if". And of course there is also Rope Trick, Mordenkainen's Magnificent Mansion, Daern's Instant Fortress, etc.
Even a module like B2 seems, in practice, to make it pretty straightforward to PCs to begin every venture into the Caves on full hit points and spells. For your long rest you pull back to the Keep - which the humanoids don't assault, at least by default - and stay there til everyone's at max hp and spells. The healing dynamic of B2, and the difference in pacing and attrition between (say) classic D&D and 4e, seems to play out at the "short rest" level, not the "long rest" level.
The Slaver modules played as an ongoing campaign, in which the PCs are operating in hostile territory on the Wild Coast and in the Pomarj, might be an example of an AD&D module which does having the "long rest" dynamics that you are pointing to. But it seems to me this makes it a bit distinctive among those classic AD&D modules.
From my point of view, I see the situation a bit differently. It's not that all-better-in-a-day healing makes these classic modules unplayable. It's that it changes their fiction. Instead of taking a month of steady action, rest and recovery to clear the ToH, it takes a week. Instead of retreating to the Keep and resting for a week or so, the PCs head out the next day (and hence the fictional constraints on changes the GM can make to the Cave environment are greater). It seems to me that this is the strongest argument for your viewpoint, but perhaps I've misunderstood you.
