nx_teacher_trainer.nx

nx_teacher_trainer.nx -- the TRAINER half of the Nishi Teacher (Teacher arc, rung R3 = spaced review). Operator: "consume all our information from the nishi teacher AND TRAINER so we get insights to improve the team's capabilities." The teacher SYNTHESIZES insights into the unified native capability store (nx_teacher_merge); the trainer SCHEDULES spaced retrieval practice over them and MEASURES coverage. Grounding = the cited learning science already in knowledge/registry/teaching_evidence.tsv (Bjork desirable difficulties: retrieval practice + spacing beat re-reading). Append-only review state in the native seg_store (tag "rev", row = {id, reps, interval, due}); current state of an id = its LAST appended row (additive, history-sacred #13). Spacing: first practice -> 1-day interval, each subsequent success DOUBLES it; an insight is DUE if never practiced OR now >= its due epoch. Coverage = (insights practiced >=1) *1000 / total -- a MEASURED readiness number. trn_practice(state, id, now) -- record a retrieval-practice event (advances spacing) trn_due_count(cap, state, now) -- how many unified insights are due for practice trn_coverage(cap, state) -- how many are under active practice (the meter) HONEST SCOPE (no-overclaim, [[feedback-no-wave-measured-exceed]]): this SCHEDULES practice and MEASURES coverage; it does NOT by itself prove the team got "smarter" -- a real capability LIFT needs the team to actually practice + a before/after task-performance measure (a further rung). license_tier: ORIGINAL