User testers (at least the kind you get on a site like usertesting.com) are very highly motivated. A paid user tester will keep at it long after a real user would bounce.
It might be a good idea to run experiments adding UI elements (and removing the walkthrough steps) for some of the gestures when you start getting a significant flow of new users.
These users were mostly random people in coffee shops, on the street, etc. I'm sure there's still a bias when someone is looking over their shoulder, but we could get a pretty good idea of whether average users 'get it' right away or floundered. We went through about 5 full redesigns of the walkthrough based on this testing before we arrived at the current version. Of course, we'll keep experimenting and improving in future versions.
It might be a good idea to run experiments adding UI elements (and removing the walkthrough steps) for some of the gestures when you start getting a significant flow of new users.