It all depends on how fast it grows.
What if the player is left without spells and hp after a hard battle
(I suppose the larian bends and remove the healing from the food).
In my opinion, this is terribly overcomplicated.
Was this directed to me about my "fatigue meter" idea? I'm not sure how fast it would grow, but Larian is pretty proud of all the analytical data they're collecting, and they might have a rough idea how many encounters they think players should have in a day, and they can figure out the distance between encounters in their game.
I have no doubt that Larian will reject my idea for being overly complicated. I'm not sure they'll find a better way to balance long and short rests, and I'm not even sure if they'll even try.
In the case of such a limited rest, fights such as the bulette or the minotaurs would be much less frequent. Instead, you have to put in simple fights with trash mobs (which are usually much less interesting).
Now how are you going to balance them? Do you assume that the player will be approaching the fight with some resources or maybe he had bad luck in the previous match and will have to rely on cantrips?
If you assume that a player has no more resources then fights can be trivial.
If you assume that the player still has some spells left, the fight may be too difficult.
Now we are adding difficulty levels on which we need to balance it.
Too easy or too difficult fights are an easy way to get bored.
In terms of design, it is a nightmare. You have to consider possible group compositions and a few other factors.
In the game, you are not able to choose next fight depending on how the player is currently doing.
This is part of the problem with adapting an attrition-based game which requires human management to a game managed by an algorithm. At the moment, Larian is using an infinite free long rest system, which removes the attrition part of the attrition based game, and that also has balance implications.