I think it is very difficult to give generally applicable advice. That is why I think a good DMG would recognise the well-known range of approaches to the came, and identify the equally well-known range of techniques suited to those various approaches.The character/player dichotomy has long been a problem for exactly the reasons presented above. In-game things should be sorted out with in-game skills and abilities, ie: if your wizard is smart he solves the thinky puzzles, not the idiot playing him. It's a great reason to use "secret" information through passing notes, that way when Bob the Wizard figures out the puzzle, even if his player doesn't, you pass him a note explaining what exactly it is he figured out.
<snip>
There are times when player action is necessary of course, but these things should be kept to a minimum. If you're trying to make what is essentially a skill challenge last an hour or more....it should either be very very complicated and difficult to roll. Simple as that. The challenge should be structured and ordered to basically only let one person go at a time, and it should be skill-limited in order to limit who can do anything at all. Honestly it's going to need to be something really darn tootin special to last upwards of an hour, especially if it is something in game that is taking under a few minutes. To great of a time differential breaks immersion, generates boredom and on the whole is confusing. None of which promote creative thinking or even slightly encourage the players to become involved.
For instance, I don't think I would ever follow the advice that you give here. Working out what actions to declare - be that in combat, in a skill challenge, in trying to solve a puzzle, etc - is the essence of play. I might kibbitz about it with the players if I feel like it, because talking with my friends about game play is fun. So, for instance, if a player is trying to work out which pathway to take for his/her PC's move action in combat, I might talk about the merits of various combinations of avoiding difficult terrain, sucking OAs, etc. But I wouldn't ever modulate such talk to reflect the mental stats or ingame features of the PC. It's table-talk, not part of action resolution or scene-framing.
It's true that clever people are less likely to make unforced errors in game play, are more likely to remember the full suite of resources available to them, etc. At my table, when someone forgets to declare some action that would have been clever to do, we call them a "bad Magic player". (Eg the player misses by 3 on an important attack, and then forgets to use Insightful Riposte to get +3 to hit as an interrupt.) If see bad Magic play in the offing, I may or may not table-talk as GM, depending on anything ranging from pacing considerations to my mood at that moment, but again I wouldn't modulate that in reference to any ingame considerations. It's table-talk.
The same thing is true in skill challenges. Is it better to suck up to the NPC, or try and bully him? That's a decision the players have to make. Do they make an Insight check to try and work out how he might respond to one or the other tactic? That's a decision for the players to make too.
The fact that clever people are likely to do better at RPGing, whereas being really good at running doesn't help even if the PC you're playing is a long-distance runner, is something that I see as an inevitable consequence of the game being an intellectual pastime - it involves words, and maths, and authorship - rather than a physical one.