I would call this a bug. The problem isn't that scale isn't scaling the handles, it's that it IS scaling the handles that have been edited, when it should not. You can check this by creating a sketch and drawing a spline, editing one of the handles, copying it, then scaling the copy. The Copy does not have the same shape as the original spline because it scaled the handle you edited. If you make the first spline fixed and put a driven dimension on the length of the edited handle, and put a driving dimension on the corresponding handle on the copy, the shape will change to the same as the original. It seems like this was just a mistake that was made in the implementation of the scale tool. I can't really think of a reason why you would want an edited handle to scale along with the spline.
Edit: I was wrong, the handles do need to be scaled by the same amount. The issue is that when you edit certain handles then scale, it does scale the handles that you edited, but its doing something else to the rest of the handles. They are not scaled by the same amount, but they are also not left the same.
Edit: I think what's happening is that Fusion is only scaling the spline data (fit points and edited handles), and then handing that to the spline solver to be re-solved from scratch instead of handing it over a known solution (scaled fit points, edited handles, implicit handles, and maybe control points) for it to start from. So I'm not so sure if its a bug or just a limitation of the spline solver, but it is a solvable problem.
