[Model Runner V2] Remove unused warmup_for_prefill method (#36762)

Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>

vllm/v1/worker/gpu/model_runner.py

@@ -532,13 +532,6 @@ class GPUModelRunner(LoRAModelRunnerMixin):
         )
         return cuda_graph_size
 
-    def warmup_for_prefill(self) -> None:
-        # For FlashInfer, we would like to execute a dummy prefill run
-        # to trigger JIT compilation.
-        if all("FLASHINFER" in b.get_name() for b in self.attn_backends.values()):
-            self._dummy_run(self.max_num_tokens, skip_attn=False)
-            torch.accelerator.synchronize()
-
     def finish_requests(self, scheduler_output: SchedulerOutput) -> None:
         finished_req_ids = scheduler_output.finished_req_ids
         preempted_req_ids = scheduler_output.preempted_req_ids