Merge nucleic/sleek-ember-seal-uady into dev

data/curation-review-v1.json

@@ -75,10 +75,11 @@
       "purpose",
       "secondary",
       "difficulty",
+      "slice",
       "review status",
       "notes"
     ],
     "generatedArtifact": "ml/purpose-classifier/.artifacts/human-review-v1.csv",
-    "completionRule": "Every sampled row must be marked accept, relabel, or reject by a human reviewer; relabel/reject decisions are applied to canonical source data before dataset-v1 can be called fully curated."
+    "completionRule": "Every sampled row must be marked accept, relabel, or reject by a human reviewer. review_data.py validates the exact sample and writes a versioned complete ledger; prepare_data.py applies relabel/reject decisions before splitting. The revised split and frozen evaluation must then be intentionally reviewed and versioned before the dataset can be called fully curated."
   }
 }

data/dataset-v1-manifest.json

@@ -9,7 +9,7 @@
     "nearDuplicateThreshold": 0.92,
     "retainedRecords": 12193,
     "reviewPath": "ml/purpose-classifier/data/curation-review-v1.json",
-    "reviewSha256": "2db763df027b545b4fd654108dcadb0717f9c9c0a63555bf8cea26b791cda370",
+    "reviewSha256": "625251a98bcdcde0bee074e3bba3c564af4347eb7955497a9c3e018aae00b30f",
     "vagueEvalPolicy": "validation/test only"
   },
   "datasetVersion": "purpose-dataset-v1",

data/semantic-audit-v1.json

@@ -53,6 +53,7 @@
       "reviewedPurpose",
       "reviewedSecondary",
       "reviewedDifficulty",
+      "reviewedSlice",
       "reviewStatus"
     ],
     "samplePurposeCounts": {