[Intel GPU] Integrate flash_mla_decode in Intel XPU attention backend (#23557)

Signed-off-by: P V R K Jyothendra Varma <polisetty.v.r.k.jyothendra.varma@intel.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
Co-authored-by: Kangyan-Zhou <zky314343421@gmail.com>
Co-authored-by: Ma Mingfei <mingfei.ma@intel.com>

test/srt/xpu/test_intel_xpu_backend.py

@@ -7,6 +7,7 @@ import unittest
 from functools import wraps
 
 from sglang.test.test_utils import (
+    DEFAULT_MODEL_NAME_FOR_TEST_FP8_WITH_MOE,
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST_BASE,
     DEFAULT_SMALL_MODEL_NAME_FOR_TEST_QWEN,
     CustomTestCase,
@@ -22,7 +23,7 @@ def intel_xpu_benchmark(
         @wraps(test_func)
         def wrapper(self):
             common_args = [
-                "--disable-radix",
+                "--disable-radix-cache",
                 "--trust-remote-code",
                 "--mem-fraction-static",
                 str(mem_fraction_static),
@@ -65,6 +66,18 @@ class TestIntelXPUBackend(CustomTestCase):
     def test_attention_backend(self):
         return DEFAULT_SMALL_MODEL_NAME_FOR_TEST_BASE
 
+    @intel_xpu_benchmark(
+        [
+            "--json-model-override-args",
+            '{"num_hidden_layers": 4}',
+            "--decode-attention-backend",
+            "intel_xpu",
+        ],
+        min_throughput=32,
+    )
+    def test_mla_decode_attention_backend(self):
+        return DEFAULT_MODEL_NAME_FOR_TEST_FP8_WITH_MOE
+
 
 if __name__ == "__main__":
     unittest.main()