nx_browser_exceed_census.nx

nx_browser_exceed_census.nx -- BROWSER-ON-ALL-OS arc, capstone: the MEASURED EXCEED census vs the incumbent browsers (Chrome / Firefox / Edge / Safari). No-wave doctrine: "s-class exceed" is CLAIMED only where measured/evidenced, and every deficit is ADMITTED. This organ turns the head-to-head into data rows, each with a verdict: A = AHEAD -- an axis where Nishi exceeds BY EVIDENCE (an on-disk artifact a mechanical probe checks) or BY CONSTRUCTION vs the incumbents' by-design facts (stated, not run). P = PARITY -- both sides do it; no exceed claimed. B = BEHIND -- the incumbents are ahead; ADMITTED (breadth debt named per row). X = NEGCTL -- liar-kill: a must-be-absent evidence file; if present, verdict RED. OVERCLAIM CONTROLS (the teeth, mirroring nx_os_exceed_gate): C2 forces the census to admit >= 6 BEHIND axes -- it mechanically CANNOT whitewash; C3 re-probes every AHEAD evidence file -- it cannot fabricate a strength; C4 the negctl. The VERDICT is census INTEGRITY, never the score. Incumbent-side facts (5-layer stacks, V8/JIT CVE class, telemetry defaults, no-NishiOS) are BY-DESIGN public facts stated as the yardstick, like dd in nx_flash_test -- NOT fabricated measurements. NEVER-BRICK (Rule 26): reads evidence, writes status artifacts only. expect_exit: 0 license_tier: ORIGINAL