It's always been the case that if you didn't have sufficient number of encounters the wizard would nova. Spells have always been intended to be a somewhat scare resource and the game has always worked better when there was pressure on the decision making about when they're best used.
One side effect of early editions lacking encounter guidelines was that the spellcasters had no idea how many encounters they might potentially face in a day. Sure, you could nova if you knew you would only have one fight in the day, or if you were terrified to the point that running out of spells was less dangerous than letting the current opponent stay alive for another few minutes; but everything was a calculated risk.
Uncertainty is as important as sheer volume of encounters when it comes to keeping the spellcasters conservative.
I'm not
@AbdulAlhazred, and am happy to be corrected if I'm getting things wrong. That said:
(1) The versatility that AbdulAlhazred is talking about is
not particularly about "encounters" or "fights" per day. To borrow a phrase from p 18 of Gygax's PHB, it's about how to "most successfully meet the challenges which [the game] poses". Everything from architectural challenges (locked doors, high walls) to environmental/geographical challenges (far distances, stormy seas) to social challenges (angry mobs, reluctant informants) to economic challenges (limited supplies, unique but powerful artefacts), etc. My experience of high-level AD&D, and of high level Rolemaster (which resemble AD&D in this respect) is that spellcasters dominate in respect of this sort of versatility. And often that's not just because of in-fiction considerations, but because of
mechanical or
system considerations.
An example concerning travel, derived from actual play experience: imagine a group of PCs whose home base is a nice, safe palace at the heart of civilisation; and whose current activity is exploring a mysterious ruin many hundreds of miles away. There is no in-fiction reason why non-spellcasters could not undertake this task: they could assemble a team of mercenaries (for safety), porters and labourers and technical specialists (to help with the travel and exploration), etc. In fact,
in my current Classic Traveller game the PCs are doing exactly this! (Of course having a starship helps with the transportation aspect of the challenge.)
But think about this from the system point of view: the need to undertake the hiring, undertake the travel, the encounter checks involved, the logistics of managing the records of supplies, hireling pay, etc. Think about the prospects that any part of this can mean that the focus of play shifts from the exploration to something else - be that a random encounter, a disloyal hireling, or some other distraction - for hours or even sessions of play. And then contrast that with the system requirements for getting everyone from home base to mysterious ruin via Teleport Without Error.
Move Earth and Disintegrate compared to digging; Charm Person compared to a mundane interrogation or seduction attempt; Detect Magic compared to engaging a sage or a bard; etc, etc. At nearly every point the system makes play easier for casters even if, within the fiction, it would be feasible for non-casters to achieve the goal in question.
Of course that's before we get to goals, like cross-planar travel or conjuring up the spirits of the dead or forging rings of power, which in the fiction are unattainable without magic.
(2) In his PHB, under the heading "Successful Adventures", Gygax emphasises the importance for players of
planning their dungeon expedition. This includes advice on mapping, on managing time spent and noise made in the dungeon, etc. Part of this of course involves making calculations about spell load (which he discusses) and spell use (which he doesn't discuss). It's not possible to know exactly how many encounters will occur on the way through the dungeon to room X, because they're on a randomised clock, but likelihoods can be calculated with a bit of a buffer being built in.
This sort of planning-oriented approach can be adopted in a wide range of play contexts beyond dungeoneering (I know this from experience - I've seen it done).
In this sort of play, casters aren't "holding back" because they don't know what's coming next. Rather, the whole point is to get the caster, spell load-out largely intact, to place/situation X, where they then deliver the spell or spells that will solve the problem. The caster is essentially the "mission specialist" while the non-casters are the transport and security personnel who make sure the mission specialist is delivered intact and good-to-go.
Which goes right back to the versatility point. The issue isn't solely or even mostly about "nova damage" - though that can also be a consideration for class balance. It's about which PCs are crucial and which are peripheral.
There are approaches to play which depart considerably from this special ops/mission-specialist/planning-oriented style. But once playing in those other ways, the whole system of spell load-outs and resource management starts to seem unnecessary and a distraction.