nx_adoptsweep.nx

nx_adoptsweep.nx -- WHAT FRACTION OF THE ADOPTER CLASS ACTUALLY MIGRATES? The D001 census (nx_drysweep, debt 1786236504) found adopter=605 of 2312 gates. But ADOPTER only means dry_apply EMITS a candidate -- it does NOT mean nx_gate_migrate verify ACCEPTS it. Proven end-to-end on exactly ONE gate (nx_activities_gate: 400 D001 -> ships, debt 1786236952), so the ACCEPTANCE RATE is 1/1 and publishing "605 migrate" would be an overclaim off a sample of one. This measures the rate. ★COMPOSES TWO RULERS AND RE-IMPLEMENTS NEITHER. The worklist is read from knowledge/status/drysweep.log -- the census is the single source of truth, not a second walk that could silently disagree -- and each gate is judged by forking nx_gate_dry_apply then nx_gate_migrate verify. This organ only TALLIES. ⚠THIS ORGAN MUTATES SOURCE. nx_gate_migrate verify COMMITS on acceptance (keeping <gate>.nx.premigrate, rule 13) and RESTORES on any divergence -- that commit-or-restore is what makes a bounded batch safe, but it is still a real edit to real gates. <max> is therefore MANDATORY and small by default: measure the rate on a batch, read the result, THEN decide the size of the next one. Never sweep 605 blind. ★NO SILENT CAPS: the DEFERRED remainder is printed; resume with <skip>. ★DURABLE AS IT GOES: one row per gate to knowledge/status/adoptsweep.log BEFORE any summary, so an interruption costs the remainder and not the run. nx_adoptsweep <max> [skip] exit: 0 all ACCEPTED | 1 at least one REFUSED/ERROR (tally still printed) | 2 usage 3 NO-CONCLUSION (census log or a ruler unresolvable) license_tier: ORIGINAL No hw writes (Rule 26).