nx_adopt_lib.nx

nx_adopt_lib.nx -- PURE core of the ADOPTION CENSUS (operator priority, seq1410). THE LAW THIS ENFORCES: **THIS ECOSYSTEM'S BOTTLENECK IS NOT BUILDING PRIMITIVES, IT IS ADOPTING THEM.** Measured three times on 2026-07-30 alone: cpu-feature-detect existed with a KAT and was registered nowhere (seq1359) · ss_close existed with 133 openers and 1 closer (seq1347) · nx_hw_cpu_count exists, is CORRECT (sched_getaffinity, cgroup-safe), and has ZERO external callers in 19,823 files (seq1410) while nx_torrent_get hardcodes MAXP=32. A build-only definition of done is why. ==> A ZERO-CALLER CAPABILITY PRIMITIVE IS A RED DEFECT, NOT A COMPLETED FEATURE. ★SCALE IS THE DESIGN PROBLEM, and the answer is the same one seq1318 taught: the naive census greps the corpus ONCE PER PRIMITIVE = O(primitives x corpus) = ~50 GB of scanning for 500 primitives. Instead we walk the corpus TWICE -- pass A collects definitions into a hash table, pass B counts every identifier against it with O(1) lookups -- so the cost is O(corpus), independent of how many primitives exist. LAW (third application today): ONE PASS OVER THE DATA, THEN ANSWER EVERY QUESTION FROM IT. Rule 15: nx_funcmine already proved this shape (bucket-chained, O(files+funcs)). Its helpers live inside an organ with a main(), so they cannot be imported -- THAT is why they are re-declared here, in a LIB, so the next consumer imports instead of copying. nx_funcmine should migrate onto this lib. license_tier: ORIGINAL Pure. No hw writes (Rule 26).