If students are familiar with the idea that fractions can be represented as repeating decimals and are introduced to π as a non-repeating decimal, then I think the idea of the set of all possibly repeating decimals should make sense to them.

> Another issue with this is that it confuses numbers with their representation.

It's probably worth acknowledging that some numbers have multiple representations just to avoid arguments about whether 0.999.. = 1.

However, is it much worse to say that a real number is an equivalence class containing 1 or 2 infinite decimals than to say that a real number is a non-empty, downwards closed, proper subset of the rationals? I agree it's nicer to define the reals as a field with some extra properties, but I'm not convinced it's wrong to say that they are their representation.