[misc] Use --cuda-graph-max-bs-decode in tests, examples, and docs (#29591)

test/manual/test_kv_events.py

@@ -46,7 +46,7 @@ class TestKvEvents(CustomTestCase):
                 '{"publisher": "zmq", "topic": "kv-events"}',
                 "--max-total-tokens",
                 32,
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 2,
                 "--enable-dp-attention",
                 "--dp-size",
@@ -190,7 +190,7 @@ class TestKvEvents(CustomTestCase):
                 '{"publisher": "zmq", "topic": "kv-events"}',
                 "--max-total-tokens",
                 64,
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 4,
                 "--enable-dp-attention",
                 "--dp-size",
@@ -327,7 +327,7 @@ class TestKvEvents(CustomTestCase):
                 "--max-running-requests",
                 4,
                 "--disable-cuda-graph",
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 4,
                 "--model-loader-extra-config",
                 '{"enable_multithread_load": true, "num_threads": 64}',