Pielorino, it seems to me that your examples of so called "undetectable" traps all commit the error of assuming that they are crafted perfectly and have no trace of the mechanism that is visible or inferrable from outside. If you add "this trap is perfect" to your premises, then obviously it will still be there in your conclusions.
Now, OTOH, the premise of most investigative fiction from Sherlock Holmes to CSI to MacGyver to the odd bits of investigation found in action movies is quite different. There is NO perfect crime, there is NO perfect trap, etc. One can cover the obvious evidence and cover the obvious evidence of covering the obvious evidence but one can neither forsee all possibilities nor always cover ALL the evidence or all of the evidence of covering all of the evidence.
The relevance to the trap-detecting scenarios is quite clear. Assuming that the trapped chest in question is a dummy--meant to lure thieves in and harm them rather than to defend anything in the chest (because if it's meant to defend something in the chest then someone needs to be able to reliably use it safely), it's still possible that some clue in the chest's construction could indicate the presence of the mist vial to the theif. Perhaps the seal is only very very nearly airtight and, over time, a little bit has leaked out, leaving a faint sent or inducing slight lightheadedness when one approaches the chest. Perhaps the wood has aged slightly differently where the pins connecting the rod to the lid are--indicating the presence of some mechanism inside the chest. Or perhaps the pins are made of a metal that stains the wood. Or perhaps, a clever observer can see that the chest has probably never been opened and everyone who went through the room seems to have given it a fairly wide berth and come to the conclusion that it's a trap. Obviously, all of these possibilities are fallible, but denying the possibility of all of them makes it an increasingly contrived "perfect crime" scenario which, while it looks good for a "what if" discussion, doesn't fit with the concept of a living campaign world and begins to be a lot more like a DM saying "rocks fall, everyone dies" than actually having a reason for rocks to fall, and a place for them to fall from. (One should note that it can also be done for any trap--even one that obviously ought to be detectable like a pit covered with some weak branches and leaves--"there might be gaps where, if one looks closely, one can see that there is no dirt underneath"-"no, the trapmaker put some dirt on the branches to disguise that", "it's hard to make a thousand leaves in one area look perfectly natural"--"no, the trapmaker disguised it perfectly as a forest floor", "maybe there are animal tracks that avoid it or an animal stepped on one of the branches and started to break the wood but jumped back just in time leaving a sign visible to the trained and careful eye"--"well, it could have happened but it didn't--in fact, several animals just barely light enough to walk across without setting it off have walked across it in the last day and the leaves, etc shifted just enough that their tracks don't give any clue as to the unnatural flexing of the ground--don't you get it, this is a perfect pit trap and it's undetectable unless you either set it off or send monsters/sheep/halflings to walk ahead of you." At this point, the ingenuity of the trap is not relevant to it's difficulty; the obstinacy of the DM is what makes it hard to find).
The idea of separate DCs 30 for finding it without risk, 15 for finding it with risk is a good one (but keep in mind that, once you've cranked the basic DC to 30, the trap is no longer CR 2)--certainly far better than "this trap is perfect; you can't find it unless you set it off." However, the whole idea of describing in detail where and how you search reminds me very much of the "well of course you didn't find the spellbook because you searched the room not the top of the table" scenario--or the scenario I played once where, right before our party climbs down a pit to retrieve a dagger, we see an ancient warning in elven scrawled in four foot tall letters along the side: "Let the evil rest here." We wasted half an hour thinking about how to get the dagger out safely (no one touches the ashes around the dagger, etc) because the DM was stuck on "you looked in the pit, not on the wall." That may be fun for a while, but I'd much rather assume that my rogue knows how to find traps and searches in the best manners he knows how than have the whole thing degenerate into a game of "I shoot you"/"But I'm wearing a bullet proof vest"/"but I'm using armor piercing bullets"/"But armor piercing bullets are less lethal so I'm still alive"/"But armor piercing bullets from a FAL are still plenty lethal--even through 2 cinderblock walls and 50 gallon water drums so you're dead"/"I would be if I'd been there but I ducked"/"I saw you ducking because I drilled a hole in the wall to look for you before I shot"/'but you drilled a hole in the wall and there's a curtain hanging from that wall so it didn't help (if you'd drilled a hole in the door it would have worked)"/"etc etc."
If that makes my preferred style of play "beer and pretzels" so be it.
