refactor: extract ensure_artifact into app/storage/artifacts.py

The artifact-upsert helper was duplicated four times (scanner.py,
table_processor.py, figure_processor.py, pipeline.py) with slightly
different signatures. Consolidates into a single keyword-only function
keyed on (document_id, storage_key) - the identity the schema already
enforces - so re-running the pipeline never creates duplicate rows.

scanner / table_processor / figure_processor now import the shared
helper directly. pipeline.py keeps a thin local wrapper to preserve
the positional call sites at three artifact upsert points (OCR_PDF,
MARKDOWN, DOCLING_JSON).

Tests: 24 passed (5 health + 19 original).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

app/ingestion/table_processor.py

@@ -7,9 +7,10 @@ import uuid
 
 from sqlalchemy import select
 
-from app.db.models import ArtifactType, DocumentArtifact, Table
+from app.db.models import ArtifactType, Table
 from app.ingestion.docling_extractor import ExtractedTable
 from app.logging_config import get_logger
+from app.storage.artifacts import ensure_artifact
 from app.storage.local_paths import key_table_json
 from app.storage.minio_client import MinioStorage
 
@@ -52,7 +53,14 @@ def persist_tables(
                 data=json.dumps(t.json_data, ensure_ascii=False).encode("utf-8"),
                 content_type="application/json",
             )
-            _ensure_artifact(db, document_id, ArtifactType.TABLE_JSON, storage.derived_bucket, key, t.page_number)
+            ensure_artifact(
+                db,
+                document_id=document_id,
+                artifact_type=ArtifactType.TABLE_JSON,
+                bucket=storage.derived_bucket,
+                key=key,
+                page_number=t.page_number,
+            )
 
         count += 1
     return count
@@ -62,23 +70,3 @@ def _summary(t: ExtractedTable) -> str:
     md = t.markdown or ""
     n_rows = max(0, sum(1 for ln in md.splitlines() if ln.startswith("|")) - 2)
     return f"Table {t.table_index} on page {t.page_number} ({n_rows} rows)."
-
-
-def _ensure_artifact(db, document_id: uuid.UUID, artifact_type: str, bucket: str, key: str, page: int | None) -> None:
-    existing = db.execute(
-        select(DocumentArtifact).where(
-            DocumentArtifact.document_id == document_id,
-            DocumentArtifact.storage_key == key,
-        )
-    ).scalar_one_or_none()
-    if existing:
-        return
-    db.add(
-        DocumentArtifact(
-            document_id=document_id,
-            artifact_type=artifact_type,
-            storage_bucket=bucket,
-            storage_key=key,
-            page_number=page,
-        )
-    )