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

test/registered/core/test_basic_sanity.py

@@ -48,7 +48,7 @@ class TestBasicSanity(
             cls.base_url,
             timeout=DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
             other_args=[
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 "4",
                 "--mem-fraction-static",
                 "0.7",

test/registered/core/test_basic_sanity_eagle3.py

@@ -70,7 +70,7 @@ class TestBasicSanityEagle3(
                 "1",
                 "--speculative-num-draft-tokens",
                 "2",
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 "4",
                 "--mem-fraction-static",
                 "0.7",

test/registered/core/test_srt_endpoint.py

@@ -48,7 +48,7 @@ class TestSRTEndpoint(CustomTestCase):
                 "--enable-custom-logit-processor",
                 "--mem-fraction-static",
                 "0.7",
-                "--cuda-graph-max-bs",
+                "--cuda-graph-max-bs-decode",
                 "8",
             ),
         )