Again, think of one of the old Sierra adventure games. You missed that item in an area you can no longer go back to, so now you're stuck. I guess you made a "strategically bad choice", LOLOLOLO
It quickly became an established norm of the genre that you grab everything that isn't nailed down. We are still talking strategy here; making imperfect decisions using imperfect knowledge, to the best of your ability, in order to overcome the challenges presented to you by the developers. The only 'bad design' that I can see in your example lies not in the fact that negligence has consequences, but that the game doesn't end with a 'loss' once the player has passed a 'point of no return', the ponr being a core element in these types of games. Suppose, that if a particular mission in an rts practically requires you to build defensively in order to counter a rush, it should by the same logic be considered bad design unless the game either tells you exactly what you need to do beforehand or makes every challenge play out in a way that allows for every possible tactic to be an equally effective counter. In Sierra adventure game terms, not having an item to solve a puzzle would be countered by a persistent tutorial telling you what needs to be done to achieve success or simply removing the need for the item, the puzzle and the solution entirely -- i.e it wouldn't be much of a game anymore.
I've been involved in these discussions before, and the most common counter-argument to brick-wall surprises is the time constraint; that making a bad strategical decision at the start of an hour long session is acceptable, while a bad strategical decision at the start of a 30-5000 hour long session is unacceptable. It's a bogus argument, because any real consequence of a combat encounter where the opponent uses surprise tactics in order to gain the upper hand, like simply hiding out of plain sight, might just as well cost you the use of valuable resources such as party members, unique items and stats necessary for survival in upcoming combat encounters, from that point onward. Of course, most of the time you can reload and try your luck until the outcome is just the one you want and any element of surprise is gone, but then again the same holds true for character creation. So if you replay the same combat encounter long enough you'll eventually match the time it takes to restart the game, but both these approaches are merely ways of trading what the player is willing to sacrifice. Are sacrifices then bad design?
There's also the possibility of dialog options leading up to consequences that are hours of play-time removed from the choice at hand, and yet I've heard no argument in favor of consequences being shallow and always following within minutes of the choice. If a brutal murder of a sleeping innocent committed by the PC might provoke a likely lethal encounter with law enforcement 40 hours later into the game, are we bound by the rules of great design to stress this fact to the player in way of discouragement? Do we spawn warning tool-tips of information, or have immortal guards spawn the second after the crime has been committed? A mere hint of danger would be little more than encouragement to a cold-blooded killer. Maybe consequences are a bad thing overall?
I say, let people find things out for themselves. If they paint themselves into corners, they can either swallow their pride and try again with what they've learned so far or make holes in the walls by hacking the game if patience runs thin. But part of the challenge of a game stems from the player's ability to react to unknown situations. Not having read the manual and having only a fair guess as to what challenges lie in wait ahead, you are confronted with ability points that need to be distributed in such a way as to aid you in your success. By making this process safe and boring, the choice becomes a cosmetic one, no more in the realm of game-play than selecting the character's appearance.