Native-O(1) Totality Benchmark — Execution-Time and Return-Value Signature of Totality vs Repair on ℝ × ℝ≥0 Carriers

Pre-registered empirical benchmark testing the Type A prediction of the characterisation theorem (Martin 2026, RSOS-260797): that carrier-level repair/maintenance surfaces at both execution time (latency variance) and return value (selection map) on implemented stacks. Amendment A2 adds parity instrumentation for all four carriers (EB, interval, affine, MC). Main claim foregrounded: EB has null parity (closed-form return), interval has non-degenerate selection parity across input sign-classes. Bundle includes pre-registration with A1+A2 amendments, benchmark source, results JSON, run log, MANIFEST.sha256, and human-readable REPORT.md.Amendment (2026-05-22): The characterisation theorem referenced here as…