The problem here is that what we want is for the Rogue to be better at using skills compared to the other classes... but not so powerful that the DM has to artificially inflate DCs just to give the Rogue a challenge. Especially considering those DCs he creates end up pushing all the other classes out of range from actually attempting them should there be a problem with the Rogue and he isn't there to do it (for whatever reason).
I say it's highly dependent on the skill.
Take a typical Rogue-ish skill: Find and Remove Traps or Open Lock.
"Picking a typical lock" is DC20. I cannot find a DC for "disarming a typical trap", but I think it's reasonable to assume DC20 as well.
This means that a 1st-level Rogue cannot automatically pick that lock or disarm that trap because her maximum ability bonus is +5 (and will always be) and her skill bonus is +3, hence take 10 is only 18. Even such "perfect" Rogue has to wait at least until level 5 to automatically pick a "typical lock" or disarm a typical trap.
On the other hand... some skills have a penalty for failure (called "hazard"). Now
that is where the take 10 features kicks in... that Rogue is not going to set off the trap and suffer the penalties thanks to her Skill Mastery. This is of course situational, clearly there are other skills such as Pick Locks where there is no hazard to avoid.
I don't know why you are worried about raising the DC if you have a Rogue in the party. I could argue that a "typical lock" is well... typical, so if you are using something worse then you are actually doing the opposite i.e. lowering the DC to let others have a better chance. It's actually ok to do either.
If you check the probabilities of success for a Rogue vs a non-Rogue, you'll notice that they are exactly the same when the DC is higher than the one corresponding to a 50% chance (i.e. the DC which requires a roll of 11). The difference is for lower DCs, when the Rogue would have at least 55% chance of success, it defaults to automatic success.
Here there is a gamestyle issue IMHO: if in your game you put easy challenges that
must be beaten in order to proceed (e.g. the locked door that is necessary to get past in order to continue with the adventure, the collapsing bridge that must be passed lest the PCs fall to death), and expect the PCs to normally beat all of them except the occasional one, then you are right that Skill Mastery can spoil that. But if you instead put hard challenges that don't block the game but enable some extras (e.g. the locked door with a bonus treasure or a shortcut behind) then they are going to have high DCs and Skill Mastery works actually great because take 10 can't help to beat, but can still keep the Rogue safe from the hazard.
Overall as I say, it's highly dependent on the skill tho... and for instance I certainly dislike this when applied to Lore checks!