Fix writeMillis access

write_result is a DataFrame and not a series.
If it was sucessful, the result should always have a single row which we access.

Author: FlorentinD

graphdatascience/query_runner/session_query_runner.py

@@ -205,7 +205,7 @@ def run_write_back() -> DataFrame:
         except Exception as e:
             # catch the case nothing was needed to write-back (empty graph)
             # once we have the Arrow Endpoints V2, we could catch by first checking the jobs summary
-            if "No entry with job id" in str(e) and gds_write_result.get("writeMillis", -1) == 0:
+            if "No entry with job id" in str(e) and gds_write_result.iloc[0].get("writeMillis", -1) == 0:
                 return gds_write_result
             raise e