Decrease the batch size of se-resnext in unittest. (#24651)

As discussed with QA, we will use p4 machine for unit test and the GPU on those machine may not have enough GPU, which can cause "test_parallel_executor_seresnext_base_gpu" failed. So I decrease the batch size.

python/paddle/fluid/tests/unittests/seresnext_net.py

@@ -173,7 +173,7 @@ model = SE_ResNeXt50Small
 
 
 def batch_size():
-    return 12
+    return 8
 
 
 def iter(use_cuda):