Add draft extend CUDA graph for Triton backend (#6705)

python/sglang/srt/speculative/eagle_worker.py

@@ -179,6 +179,7 @@ class EAGLEWorker(TpModelWorker):
             self.has_prefill_wrapper_verify = True
         elif self.server_args.attention_backend == "triton":
             from sglang.srt.layers.attention.triton_backend import (
+                TritonAttnBackend,
                 TritonMultiStepDraftBackend,
             )
 
@@ -187,7 +188,10 @@ class EAGLEWorker(TpModelWorker):
                 self.topk,
                 self.speculative_num_steps,
             )
-            self.draft_extend_attn_backend = None
+            self.draft_extend_attn_backend = TritonAttnBackend(
+                self.draft_model_runner,
+                skip_prefill=False,
+            )
             self.padded_static_len = self.speculative_num_steps + 1
             self.has_prefill_wrapper_verify = False
         elif self.server_args.attention_backend == "fa3":