I've taught at the U.S. Physics Olympiad's training camp, and also tutor high school students who want to learn a ton of physics quickly. (These kids usually find me themselves, through Google.)

My most obnoxiously smart student was this 12 year old who would interrupt my lectures on classical mechanics with tricky questions I couldn't solve. The next year he came back, doing the same thing but for quantum mechanics, and the year after that he got bored of this and enrolled in an elite college. I want to stress that none of this required great wealth or privilege. The cost for his books was a small fraction of what other middle class parents would pay for, say, music lessons or SAT tutoring.

But a parent has to sign the permission slip for training camp, or drive the kid to tutoring. It's harder (not impossible of course) when the parent is abusive or antagonistic.

Any examples of those questions?

When covering Huygen's principle: the wavelets always give a backwards-moving wave in addition to the forwards-moving wave. In other words, Huygen's principle is time symmetric. So in real life, why don't you get the backwards wave?

When covering the principle of least action: often, applying it will give infinitely many solutions, or none at all. An example with no solutions would be the harmonic oscillator with x(0) = 0, x(t) = 0, and t not a multiple of half the period. An example with infinitely many solutions would be the principle of least time applied between the two foci of an ellipse with reflecting walls. So what happens in these cases? Does Lagrangian mechanics just not work?