Walk dimension

• 2026-08-28 [Verified] Mass does not fix music: codes 127 and 239 share fill 7 and dimension log(7)/log(3) = 1.771244 with identical density at every level, yet their walk dimensions separate by about 0.39 (2.26 against 2.64), both generators agreeing and every drift bar an order of magnitude smaller than the gap. Witness: lab/rs/walk-dimension.
• 2026-08-28 [Conjecture] Fixed mass does not fix the drum either: all nine fill-8 tiles recomputed at side = 243 and side = 729 under both boundary conventions give centre 1.2111, edge-midpoint 0.9661, corner 0.9634 (shipped convention 1.2093 / 0.9666 / 0.9633) at full held fraction, so the 0.25 spread is not a dropped-component artifact and "mass falls faster than stiffness" fails at exactly fixed mass.