Harden split background promotion preflight

scripts/README.md

@@ -724,6 +724,11 @@ presence, required `pure_empty_negative` and `sparse_building_context`
 background categories, Python/curl availability and the frontend API proxy
 envelope without starting inference:
 
+For older operator manifests that predate explicit `background_category`,
+preflight uses the same fallback as the matrix runner: background samples with
+`reference_feature_count == 0` are treated as `pure_empty_negative`, and
+background samples with references are treated as `sparse_building_context`.
+
 ```bash
 PROMOTION_POSITIVE_PORTFOLIO_PATH=/mnt/user/appdata/geointel/artifacts/detection-quality-matrix/multi-sample/aoi1024bg512r3e50-positive/multi_sample_quality_summary.json \
 OPERATOR_SAMPLE_MANIFEST_PATH=storage/operator-data/operator_samples_manifest.json \

scripts/run_split_background_promotion_workflow.sh

@@ -110,11 +110,24 @@ samples = payload.get("samples") or payload.get("items") or []
 if not isinstance(samples, list) or not samples:
     raise SystemExit(f"Operator sample manifest has no samples: {manifest_path}")
 
-background_categories = {
-    str(sample.get("background_category") or "")
-    for sample in samples
-    if str(sample.get("sample_role") or "") == "background_candidate"
-}
+background_categories = set()
+for sample in samples:
+    if str(sample.get("sample_role") or "") != "background_candidate":
+        continue
+
+    category = str(sample.get("background_category") or "").strip()
+    if not category:
+        try:
+            reference_feature_count = int(sample.get("reference_feature_count") or 0)
+        except (TypeError, ValueError):
+            reference_feature_count = 0
+        category = (
+            "pure_empty_negative"
+            if reference_feature_count == 0
+            else "sparse_building_context"
+        )
+
+    background_categories.add(category)
 missing = {"pure_empty_negative", "sparse_building_context"} - background_categories
 if missing:
     raise SystemExit(