I'd use a short extension as possible on a torque wrench for the same reason I use a short extension normally - control and "wobble". If you've ever had to use two feet of extension and a big bar to loosen a bolt you know what I mean - it's hard to apply a gentle, accurate force. Not the kind of thing you want to do when putting the nuts on your connecting rods.
What goes in must come out - if you apply 100 foot-pounds of force at the end of your bar, it's going somewhere. In the perfect world, that would all be applied evenly on the nut or bolt as 100 foot-pounds. In the real world, the threads are dirty, your universal joint is flexing and your cheap 1/4" extension is twisting and absorbing some of that 100 foot-pounds. I have a 1/2-3/8" adapter that twisted the "head" off into the socket under load - how many bolts did I torque down incorrectly because the adapter was tearing apart? Like I said, for lugnuts where most people just go by feel, even 10% accuracy is probably close enough, but for stuff like connecting rod bolts and head bolts you best be following the proper procedures.
So, no, using an extension won't automatically cause the torque values to be wrong. You're not gaining or losing any mechanical advantage. What you are risking is loss of control while trying to be accurate while holding all the bars and the possibility of "frictional losses" - an example if you have cheap (or undersized) tools that are flexing and absorbing the torque. (I'm not sure if frictional losses is the right word.)
Ray