Add Arm64 CPU Phase 1A CI bootstrap (#22123)

Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>

test/srt/cpu/test_server_args_backend.py

@@ -0,0 +1,35 @@
+import unittest
+from unittest.mock import patch
+
+from sglang.srt.server_args import ServerArgs
+
+
+class TestServerArgsCPUBackend(unittest.TestCase):
+    def _make_server_args(self, attention_backend=None):
+        server_args = ServerArgs.__new__(ServerArgs)
+        server_args.device = "cpu"
+        server_args.attention_backend = attention_backend
+        server_args.sampling_backend = None
+        return server_args
+
+    @patch("sglang.srt.server_args.is_host_cpu_arm64", return_value=True)
+    def test_arm_cpu_defaults_to_torch_native(self, _mock_is_arm64):
+        server_args = self._make_server_args()
+
+        ServerArgs._handle_cpu_backends(server_args)
+
+        self.assertEqual(server_args.attention_backend, "torch_native")
+        self.assertEqual(server_args.sampling_backend, "pytorch")
+
+    @patch("sglang.srt.server_args.is_host_cpu_arm64", return_value=False)
+    def test_x86_cpu_defaults_to_intel_amx(self, _mock_is_arm64):
+        server_args = self._make_server_args()
+
+        ServerArgs._handle_cpu_backends(server_args)
+
+        self.assertEqual(server_args.attention_backend, "intel_amx")
+        self.assertEqual(server_args.sampling_backend, "pytorch")
+
+
+if __name__ == "__main__":
+    unittest.main()