I believe that good math courses are structured around measurable learning goals. For example, "can correctly replace a line integral with an equal double integral using Green's Theorem" or "can use the Sylow Subgroup Theorems to prove the smallest non-abelian simple group has order 60."
I also believe that good math courses should give students a deeper understanding than "follow these rules". In particular, at least the major level courses should certainly include proofs of main theorems, such as the Sylow Subgroup Theorems.
But I cannot come up with a measurable learning goal (one that doesn't start with "understands") that would motivate a class activity that includes proving important theorems.
My example of using the Sylow Theorems is a good learning goal, but if that is the only goal, there is no reason to explain the proof of the Sylow Theorems. But I also feel that "Can reproduce a proof of the Sylow Theorems" is going too far.
I hope you can understand the tension I'm getting at. Have you found any way to justify teaching well-known (beautiful!) proofs in a learning-goal motivated course design?