Description
D-LBL5. Bump RUBRIC_VERSION; per-stack dimension = weight-averaged labelling score over signals with a published score, withheld when none; re-derive maturity view inputs; CHANGELOG note on the one-time shift.
Acceptance Criteria
- #1 RUBRIC_VERSION is bumped and the dimension is withheld when no signal publishes a score
- #2 Maturity tests are updated with the changed intent named in the task notes
Definition of Done
- #1 just test
- #2 just tf-validate
- #3 just check-identifiers and just no-em-dashes both return clean
Implementation Notes
loop18: attempts0, review-repair rounds0; no source files touched. D-LBL5 prerequisitedevAC1 is false after exact measuredT2 primarycoverage refusal. Resume after GCI-0119.15 (prove labelling live on dev) AC1 accepted; keep RUBRIC_VERSION, existing cardinality dimension and derived input ownership unchanged until then.
loop19 (main thread): landed 317465c after dev AC #1 accepted on v0.9.1 (R-dev8). Owner decision 2026-10-08: label_inventory disabled by config -> maturity/value views and metrics publish without the dimension; enabled but unsatisfied -> withheld/absent (hydrate.DISABLED_OK, hydrate.filter_metrics, also applied before T3 carry save). Key cardinality_discipline kept; RUBRIC_VERSION 2. Changed test intent: removed three cardinality-ratio tests (scorer removed); added withheld-when-none, weighted-average, applicable-not-evaluated weight, compose-integration, rubric floor text, hydrate disabled/unavailable and scan metric-absence tests; dashboard version test pins RUBRIC_VERSION. VIEW_INPUTS re-derived: maturity, maturity_dimensions, maturity_summary, value_benchmarks, value_summary gain label_inventory (synthetic fixture extended via bin/make_compose_fixture.py –synthetic-label-scores). Independent review two rounds (accept), CodeRabbit 0 then 1 minor (carry ordering, fixed). just check on main: 2444 passed, 2 skipped. Not yet released or deployed.