And here, I was convinced that clerics had heavy armor because they were supposed to get in there and heal, where all healing spells were limited to touch range.
I can only report how, in my experience, the B/X cleric and (traditional) AD&D cleric played. In combat healing was not the norm, because there was generally no need for it. Combat did not routinely reduce hit points to life-threatening levels, and the hit point pool was more of a "per day" resource than a "per encounter" resource.
That's not to say that hit point reocvery wasn't useful, but it was generally underaken between combats, to top up the pool for the next combat, rather than during combat to stop someone dying.
From what I've gathered, that was something which came rather late to the edition
As [MENTION=336]D'karr[/MENTION] has pointed out, Divine Power dates from July 2009. I think that's fairly early in the edition.
it's pretty much a given that both sides will be using the same healing rules, so any change to those rules would affect both sides evenly.
I'm not entirely sure what you mean by this, but NPCs and monsters in 4e largely do not heal in any mechanical fashion, and hence as a general rule don't use the same healing rules. Rather, they have bigger pools of hit points.
This feeds into the next point:
it's weird that 4E would keep the in-combat healing and remove the action component, when doing so would make combats drag on so much longer. It seems like they could have solved a lot of problems by just re-branding all heal spells as out-of-combat rituals.
I don't fully follow this, but to the extent that I do it seems to get 4e more-or-less backwards.
There is relatively little out-of-combat healing magic in 4e. Out-of-combat healing is a matter of spending healing surges. There are effects that can buff this - the bard's song of rest, a cleric using an encounter healing power during a short rest, etc - but the kernel of the healing is the spending of a surge. There is some surgeless healing (eg cleric daily Cure spells) and some surge recovery magic, but it is relatively rare; surges are the "hard cap" on the amount of adventuring that can be undertaken between extended rests.
In-combat healing, on the other hand, is fundamental to the play of 4e. There are two aspects to this, one connected to combat encounter building, the other to combat encounter pacing.
With respect to
building: the out-of-combat healing rules ensure that, in nearly all cases, PCs will enter combat at full hit points. Monsters are built around this assumption, and the encounter difficulty guidelines in turn reflct this assumption about monster design. One feature of 4e is that monsters do relatively more damage than their equivalents in B/X or AD&D (except at 1st level, where 4e is far less swingy than its predecessors). It is expected that combat will deplete a large amount of many characters' hit points.
With respect to
pacing: this hit point depletion will, typically, be
noticed by the players. They will feel their PCs under pressure, and try and respond. This can involve a range of mechanical options, but one key one is to unlock healing surges during combat, thereby mitigating that damage. This is where in-combat healing comes in: its principal function is to enable healng surges to be used outside the context of a short rest. That is, it is about
unlocing a source of healing rather than, itself, being a source of healing.
The practical effect of this, especially when contrasted with NPCs/monsters larger hp pools but lack of healing surges and the ability to unlock them, is to produce a fairly standard combat dynamic: the PCs (and thereby the players) feel the force of the NPC/monster assault, as they lose hit points at a (proportionately) faster rate than their enemies. But the PCs have a greater depth of resrouces to draw on - unlocking healing surges to mitigate damage, and deploying encounter and daily powers to step up their output relative to that of their enemies - and hence the tide turns.
As one example, the 28th level defenders in my 4e game have around 200 hp and well over a dozen surges each. When they spend a surge they may regain anywhere from 50 to 70+ hp (depending on character, on riders on cleric spells, etc). It is not especially uncommon for one of them to spend 4 or more surges in a combat, as they are pushed towards unconsicousness and death by their enemies but then rally either under their own steam (second wind) or at the urging of the party's cleric-ranger (Healing Word, Word of Vigour) or sorcerer/bard (1x/day Majestic Word).
As far as design is concerned, this is pretty much the opposite of the classic D&D attrition model of hit point loss, where a typical encounter depletes only a modest amount of a daily hit point pool, and where the main function of clerical healing is to extend the size of that pool on a per-day rather than per-encounter basis.
Whether it is a good design or not is of course disputed. But changing it in the way you suggest - ie going back to the traditional D&D model of limited relevance to in-combat heaing and an emphasis on magical healing as the per-day resource - would be a fundamental change. The whole healing surge subsystem, for instance, would need reworking from the ground up; and the removal of in-combat heallng would eliminate the pacing aspect that is central to 4e combat design.
And on the matter of pacing: once you decide to make the "pushed to the ropes and bounce back" approach central to your combat pacing, you are already going to have long combats, almost inevitaby, as combats will have to last for four or more rounds if they are to deliver that experience (say, 2 rounds of being pushed to the limit, a round of resurgence, and then a round of cleaning up). In my personal experience many 4e combats, especially at paragon and above, last more than 4 rounds.
In this context, introudcing even more healing and thereby reducing PC damage output risks making combats even longer. Whereas the default 4e approach of healing being a complementary function helps maintain a minimum PC damage output. I think the rationale for that default approach is pretty clear, whether or not one actually likes the result.