Dedup ids at load; prep v3 corpus swap

app.py

@@ -111,6 +111,7 @@ def load_table(cfg: str, table: str, has_param: bool):
             SELECT *, embedding[1:{EMB_DIM}] AS s
             FROM read_parquet({plist})
             WHERE embedding IS NOT NULL
+            QUALIFY row_number() OVER (PARTITION BY id ORDER BY last_modified DESC NULLS LAST) = 1
           )
         )
     """)