[CUDA] try to abate some flakiness in test_stream_event_nogil (pytorch#148796)

threshold twiddling as one in a few dozen runs tend to fail the current threshold

Pull Request resolved: pytorch#148796
Approved by: https://github.com/Skylion007

Author: eqy

test/test_cuda_multigpu.py

@@ -815,8 +815,8 @@ def test_stream_event_nogil(self):
             # it may vary on different hardware in different environments.
             # Therefore, this test uses relative comparisons, checking if the
             # sum of parent and child threads execution time is greater than the
-            # real execution time by least 40%.
-            self.assertGreater(parent_time + child_time, total_time * 1.4)
+            # real execution time by least 30%.
+            self.assertGreater(parent_time + child_time, total_time * 1.3)
 
     # This test is flaky for ROCm, see issue #62602
     @skipIfRocm