[FLINK-26543][python] Fix the issue that exceptions generated during …

…startup are lost in Python loopback mode

This closes apache#19013.

flink-python/pyflink/fn_execution/beam/beam_boot.py

@@ -86,7 +86,9 @@ def check_not_empty(check_str, error_message):
                 worker_id=worker_id,
                 provision_endpoint=ApiServiceDescriptor(url=provision_endpoint),
                 params=params)
-            client.StartWorker(request)
+            response = client.StartWorker(request)
+            if response.error:
+                raise RuntimeError("Error starting worker: %s" % response.error)
     else:
         logging.info("Starting up Python harness in a standalone process.")
         metadata = [("worker_id", worker_id)]