This is an empirical claim, and there's a wide range of RPGs against which we can test it. And the outcomes of that test will probably be different for different testers, who have different preferences.You don't need a TPK situation to make players feel like their resources are being tested.
Straight healing is one of the least interesting ways of handling player damage. Debuffs, buffs, strategic positioning, bottle necks, scenery, just 'not getting hurt at all' are all more interesting way to get players to deal with it.
The thing about herbalism that I don't like is that it use campaign based ressources (gold) to heal. This create situations where if you have unlimited cash you can heal to full when you want but if you don't have cash it's not much healing (it's easier to wait). The solution is to nearly remove all the gold cost of herbalism but limit it by day (assuming we stay within a daily paradigm of ressources). You can just say that these potions need the power of the alechemist to work and he can only make 2 per day maximum (or 1, depending on balance, etc). If you make a 3rd one, the first stop working, etc. Then the alchemist feat is basically allowing someone to be an off-healer and reduce the burden on having a cleric in theory.
This is an empirical claim, and there's a wide range of RPGs against which we can test it. And the outcomes of that test will probably be different for different testers, who have different preferences.
For my own part, I think active defence is interesting in a system like Rolemaster or HARP (where you allocate attack and parry each round from a common pool) but less interesting in Runequest (where each PC has a fixed attack percentage and a fixed parry percentage). 4e has persuaded me that PCs being hit is often more interesting than PCs being constantly missed, because a result of being hit is movement, conditions etc which change or raise the stakes. But a system whose dynamics rely on the PCs being hit will probably need in-combat healing (as 4e generally does).
In my view, the big problem with attrition as the stakes of combat is that (i) it is boring until the last fight is actually taking place, and (ii) it can fail to generate the requisite stakes if the players get to decide whether or not their PCs engage in any given combat.
I can think of at least one way to reconcile no in-combat healing with dynamic combat - design a system in which NPCs, monsters and PCs have a reason to use (non-damaging) pushes, grapples etc instead of damaging attacks. (Burning Wheel has a degree of this to its combat system.) But that would be something of a change to D&D combat.

(Dungeons & Dragons)
Rulebook featuring "high magic" options, including a host of new spells.