Yes Remi, generally you are right. One point though:
At the end of such a method, there is only one additional step to do: Calculate the resulting Spiro points starting from the transformed Bezier path. It's simply the reverse operation of the normal Spiro algorithm.
I see no need in such a reverse operation. If I get you right, you need this only to apply transformations to the Spiro control points. But this can be done staight forward. The shape is rendered as matrix transformed bezier, and control points are simply transformed prior to rendering.
So it's even simpler.