feat(scripts): ShmuggingFace preview site builder — v1.0.2, hardened (PR 8.4)

.agent-plan.md

@@ -113,7 +113,7 @@ _Source: `docs/external_review/summaries/v1_release_review_synthesis.md` — cro
   - Labels: `type: test`, `layer: validation`, `layer: render`
   - Size: S (~450 lines, mostly tests)
 
-- [ ] **PR 8.4** — `feat(scripts): integration script + preview hardening`
+- [x] **PR 8.4** — `feat(scripts): integration script + preview hardening`
   - **Regenerate lockfile + bump to v1.0.1** (HIGH): delete `package-lock.json`, update `package.json` pin to `github:ShmuggingFace/ShmuggingFaceCore#v1.0.1`, regenerate via HTTPS. Fixes SSH lockfile and gets the socks/laundry copy fix in one step.
   - **Remove fabricated Kaggle usability scores and medals** (HIGH): delete `TIER_USABILITY`, `TIER_MEDAL` constants from `build_shmuggingface_site.py`. These are dead config today (the framework ignores them) but latent misinformation.
   - **Remaining preview-generator cleanup** (HIGH): `build_shmuggingface_site.py` no longer exists on `main`; if a ShmuggingFaceCore build path is reintroduced, it must consume or validate against the canonical lint gate from SMF-PR5 rather than hard-code task, tags, license, splits, or schema.

leadforge/validation/invariants.py

@@ -25,6 +25,13 @@
     SNAPSHOT_FILTERED_TABLES,
 )
 
+# Feature-dictionary rows that are intentionally present only in the student
+# bundle and are exempt from the subset check in check_exposure_monotonicity.
+# ``split`` documents the partition-label column that exists only in the flat
+# ``lead_scoring.csv`` convenience export (student_public only); the
+# instructor bundle's feature dictionary covers the raw Parquet columns.
+STUDENT_ONLY_DICT_ROWS: frozenset[str] = frozenset({"split"})
+
 
 def check_determinism(bundle_a: Path, bundle_b: Path) -> list[str]:
     """Compare two bundles that should be identical (same seed/config).
@@ -169,12 +176,13 @@ def check_exposure_monotonicity(student_bundle: Path, instructor_bundle: Path) -
 
     # feature_dictionary.csv: student rows must be a subset of instructor rows
     # (by ``name``).  For names present in both, the metadata must agree.
+    #
     s_dict = student_bundle / "feature_dictionary.csv"
     i_dict = instructor_bundle / "feature_dictionary.csv"
     if s_dict.exists() and i_dict.exists():
         s_df = pd.read_csv(s_dict).set_index("name")
         i_df = pd.read_csv(i_dict).set_index("name")
-        extra_in_student = set(s_df.index) - set(i_df.index)
+        extra_in_student = set(s_df.index) - set(i_df.index) - STUDENT_ONLY_DICT_ROWS
         if extra_in_student:
             errors.append(
                 "feature_dictionary.csv: student has rows missing from instructor: "