Except for the very many times when what makes the most sense is a continuation of the status quo, i.e. nothing happens, which isn't allowed by said rules.
And right then is where the overarching question of rules-first or fiction-first rears its head.
So there are a few ways to look at this.
First, have all those factors actually been set already? The weather and the location of people within the location and all that? In some games... even some games of D&D... that very likely may not be the case.
Second, even if all those factors have been established... it's still pretty easy to come up with something. The thief breaks his lock picks, and will now have to find another route into the place without use of his lock picks. Or, he spends so much time on it, that someone does come along... a guard, or maybe just lovers out for a tryst. These don't violate any kind of fiction-first approach. They actually help make the setting feel like its alive and moving and things beyond the PC are happening, which is a huge part of what everyone was insisting is vital to play.
Third, if nothing will happen if you fail... if there is no consequence... some people don't pick up the dice. If there are no time constraints, no risk of being discovered (which to me seems inherent to breaking in to a location, but whatever)... then why bother with a roll? Why not just let them break in? Doesn't seem to be any kind of violation of fiction first to let a thief pick a lock. In fact, making them roll despite the lack of risk seems more like a rules-first approach. It seems beholden to the idea that to succeed at something, you have to roll. But most games allow for the GM to decide a roll is not needed.
It's not just that the stakes have been set, but more so that the lockpicking attempt has nothing to do with the cook. Being discretely different events means that there should be different rolls to determine each.
Why would it have nothing to do with the cook? If you're breaking into a kitchen, and there's a cook there, it seems to me that the things are connected.
Also, there's no reason that a game cannot use conflict resolution instead of task resolution. Like we look at the entire situation... breaking into the kitchen and a cook being there... and then the player's roll tells us how the situation is resolved. A high roll means successfully, a low roll means unsuccessfully. On the unsuccessful result, what is the outcome? Maybe they can't even pick the lock, or maybe they do get in, but the cook notices.
There's no reason that we MUST use task resolution. Some games specifically use another approach.
Another thing is the level of realism, which may or may not be an issue to any particular traditional DM. For me it is. In order for the cook to be there in the dead of night, the cook would have to have woken up in the middle of the night for some reason, decided to get up rather than try to go back to sleep, then have that wake-up happen to be on the night that the PC is trying to pick the lock, and lastly have the timing be during the brief period where the PC is picking the lock.
Or perhaps as a servant, they must wake up super early to get the stoves going and start to prepare breakfast? Depending on the location... and one that has a dedicated cook sounds to me like a manor or a castle or a place of affluence (as does breaking in). As such, there are likely servants active at all kinds of hours.
Seriously... this inability to make fail forward work is kind of remarkable. People are actively using their imagination to make it not work. All you have to do is flip that. Use your imagination to make it work. The idea that one of these things is problematic and the other isn't, is very strange.