Try it and you will be impressed. Artifacts generated by MP3 compression will be far worse than any such time remapping artifacts.
That's a fatuous argument. It depends on the amount of time remapping and the level of MP3 compression. For example, it's not difficult to apply too much time stretch/compression resulting in completely unrecognisable audio.
Unless something dramatic has changed with time compression/expansion software in the few days since I used it last, I really don't need to try it. I've been using time compression/expansion software by Serato, WaveMechanics, Melodyne, Waves and DigiDesign for over a dozen years, you think I'll "be impressed" by trying it again?
In US, we lovingly call artifact seekers as "Measurebator"s.
That's a coincidence, in the UK we not so lovingly call filmmakers who don't care that their sound is distorted or out of sync "Wankers" (a colloquialism for masturbater), small world!!
I'm not saying that drift problems are incurable. Using time compression/expansion software, along with other editing tools and methods, this problem can be solved but it's usually tricky, time consuming and doesn't give particularly accurate results. It's certainly a potentially serious problem to consider if the OP is thinking of hacking his/her DSLR to provide more than 12mins continuous recording time.
We never run into this issue in our narrative work because realistically a LONG take is maybe 5 minutes. I hear my audio guy talk about it however.
After 5 mins you are certainly getting drift but it's probably too small an amount to be noticeable. I can understand why your audio guy talks about it though. It might make an interesting test, if you've got nothing better to do one day: Leave the camera recording for 12 mins and use a clapper board every 10 secs or so, then import the video (and sound) into your NLE and see how much drift you're getting. At least you'll know if the drift is noticeable within the 12min limit and if so, at what point it becomes noticeable.
G