The full thing (for me) might be, "Goal and approach, followed by dice rolls only in the case of uncertain outcomes AND meaningful consequences, of which the player should be aware before committing to a roll."
And then the question becomes: because dice rolling with risk/reward is fun, how can we take some of the cases where the consequences aren't obvious and create fun challenges?
My first comment is "Why are you stuck on meaningful consequences of failure?" as the sole means of achieving the goal of fun meaningful consequences and fun challenges?
Imagine we reverse the argument and insisted that the only valid challenges were those with meaningful consequences of success. That is to say, if we go back to your original post and look at questions like: "One question that might arise is whether failing a stealth check, and thus failing to hide, really counts as a consequence. Isn't that the same outcome as not rolling at all?", and looked at those questions from this new perspective they are pretty baffling. If we only look at meaningful consequence of success, the worry about stealth checks goes away, because it's obvious that success on a stealth check involves a meaningful consequence of success - you get to stab the sleeping dragon. Thus, while failure on the success check is the same as not trying to be stealthy, success on the stealth check is definitely not the same as not trying to be stealthy.
This hypothetical poster insisting that we always need a meaningful consequence of success before the challenge is of the right sort, and not examining meaningful consequences of failure, probably would have asked a completely different sort of puzzle and assigned it medium difficulty.
To wit something like, "Suppose a character is shoved toward a precipice and must now make some sort of check to avoid going off. It's not at all obvious that there is a meaningful consequence of success. After all, succeeding in this case just maintains the status quo as if the challenge had never happened. Perhaps the person should gain some sort of advantage for having succeeded, but it's not obvious what that advantage should be?"
From your perspective on focusing on the meaningful consequences of failure, that example probably seems pretty silly. The PC has avoided going off a precipice. There was risk involved. There were meaningful consequences of failure.
But if rolls with no meaningful consequence of success are good and valid, why aren't rolls with no meaningful consequence of failure good and valid provided that there is some meaningful consequence of success? Why aren't the two symmetrical?
Back in the real world, you can probably think of many cases of difficulties where failure had no meaningful consequences, difficulties where success meant maintaining the status quo but failure was avoided, difficulties were failure meant maintaining the status quo but success was avoided, and difficulties where success or failure meant the status quo was changing either way.
I think the answer to your question remains, there is no answer to your question in the general case and in the way you've phrased it, particularly if you are going to preemptively restrict the sort of answers you get to the sort that you assume will solve the problem. Imposing worse than do nothing consequences on every situation is going to either break verisimilitude, or be antagonistic in practice, or be railroading in practice. Retroactively trying to transform the scenario into one where whatever the PC's propose it has a worse than do nothing consequence isn't a solution to every problem, and may indeed make things worse. What I would encourage you do to is try to proactively set up situations so that there is some sort of fun consequences in playing out the scenario, and when you find yourself outside of that, to move through it as quickly as won't feel like a choo-choo train to get to the "good parts".
