For the A4-versus-Letter question, I would keep marker_width_mm fixed and treat the surrounding page layout as the variable. Two paper layouts could carry the same physically sized marker, with a printed calibration line outside the recognition area so a parent can detect unintended fit-to-page scaling with a ruler.
That gives troubleshooting a more concrete starting point than the paper-size label alone. A useful acceptance check would print both layouts using actual-size and fit settings, then measure the marker and test recognition. It would separate an incorrect physical scale from a layout that merely has different margins.