fix flash attn check

Signed-off-by: jiqing-feng <jiqing.feng@intel.com>

Author: jiqing-feng

optimum/exporters/ipex/modeling_utils.py

@@ -628,7 +628,7 @@ def postprocess_attention_output(self, attn_output):
         return attn_output
 
     # Maybe removed after torch 2.6 released
-    def has_flash_attn(query):
+    def has_flash_attn(self, query):
         if query.device.type == "cpu":
             return is_torch_version(">", "2.4.99")
         elif query.device.type == "xpu":