sort mutates source.

So to avoid side effects, it must be copied before use.

If the kata wanted a CPU efficient solution it would have run very large data sets as part of the final tests.

Clever - Maybe because your solution was the obvious one and this has fewer lines of code and leverages already available methods?

Best practice? Eh, maybe not, if I were using large data sets this obviously isnt ideal, but for small values of n like this challenge, it doesn't make a lick of difference.

If I wanted something more efficient I would have used a different approach, but it's a kata, not a time sensitive app.

I don't see how this could be considered "Best Practice" or "Clever".
It needs O(n*logn) in time to do something that could be done in O(n)