Because toughness, in the fiction, is a constant: if it takes 35 points worth of damage for a merchant or a wolf or even another ghoul to kill Bob the Ghoul, that tells me it takes 35 points of damage for anything to kill that same ghoul, because that's how tough that ghoul is.
That's really not "in the fiction," though, that's /in the system/.
In fiction, a creature that the hero has a hard time beating down, one time, might go down quickly, another. And, IRL, randomness of terminal ballistics and the remarkable resilience and frightening fragility of human life is much, much stranger than fiction.
If they're already designing all their spells-powers as 'new' then the field's wide open to on-the-fly design whatever the frick we like, as this rationale immediately removes any requirement to stick to what's in the PH or other sourcebook(s). I'm not sure this would be viable in any edition.
Page 42 could be interpreted that way, if you like. But it's very clear, in 4e, that you can 'fluff' a spell however you like so long as it doesn't cross the line of changing the mechanics. You don't /need/ a new arcane power to be published or a vague DM-fiat procedure to create a new mechanic in order to get a 'new' spell, /in the fiction/. You just take your idea for a new spell, pick an existing one with mechanics that fit, and re-skin it to match.
It's the same thing 3e did with weapons after trimming the list so heavily and - with the glaring exception of the Katana - that worked just fine. (Heck, voluminous as the 1e weapon list was, it /still/ used re-skinning equivalency.)
In the fiction it ought to be trivially easy for a DM to find a way to explain why a new spell failed, particularly if the player has the PC go and ask someone and-or has their PC do their research and design while in contact with other wizards. In fact I'd say a DM who doesn't explain why it went wrong is shortchanging the PC/player.
IIRC, the 1e spell-research rules specifically said the player wouldn't know whether he failed in his research because the DM deemed the spell impossible (unacceptable) or because he just got unlucky. Of course, it's been a while...
Perhaps, though a DM/table can choose to ignore resource management in any edition should they so desire - nothing special about 4e in this regard. That said, doing so would have different knock-on effects in each edition and maybe these are fewer or more subdued in 4e?
Not ignore in the sense of removing the resource restrictions, just shift the focus away from. That is, in 4e, if you shift the pacing of play away from challenging PCs on a resource-attrition schedule, the classes remain balanced & contributing alongside eachother, and only the relative difficulty of encounters and other challenges is impacted. In any other edition, deviating too much from expected pacing quickly makes resource-heavy classes overpowered - or, on the other extreme, overextended - compared to the resource-light classes, and the dynamic of play becomes uneven, with some players wondering why they even show up.
I see the game-play efficiency rationale but to me the internal consistency is paramount; and when it conflicts with efficiency, efficiency just has to take a back seat.
Just remember that internal consistency is internal /to the fiction/, not the system.
Oh, I'll roll all 20 of those - if only because my game has crits and fumbles - and I'll also expect the player to roll for each attack vs. the mooks.
That's fine for you. 1e didn't have crits or fumbles, and did recommend just 'taking the average' to save yourself rolling all those unlikely-to-hit/unlikely-to-miss attacks. So the precedent for alternate resolution is there.
That, and in the 4e adventure modules I've seen (and run!) it's rare that the design calls for hordes of minions - more common seems to be that there's maybe one minion for each non-minion in a given encounter, which makes the too-much-rolling issue a moot point.
You can have dozens of minions in a high level encounter, and they're quicker & simpler to deal with than dozens of wildly under-leveled monsters, while staying more relevant to the encounter. That's the point, and it works pretty well.
Classic D&D had a similar point - with fighters 1/level attacks, taking averages, and even falling back on chainmail (or later Battlesystem) - but successive eds were looking for better ways precisely because that didn't work so well. 4e found one. 5e tried something a little different (not /that/ different, for instance, all 5e monsters have a don't-roll-damage option like 4e minions) - BA, and TBH, it retains too many of the original issues, and introduces a new one: being outnumbered telling too heavily.