[Bugfix][Spec Decode] Avoid double call of Ngram CPU (#36952)

Signed-off-by: Ekagra Ranjan <3116519+ekagra-ranjan@users.noreply.github.com>

vllm/v1/worker/gpu_model_runner.py

@@ -4247,15 +4247,6 @@ class GPUModelRunner(
                 self.input_batch.token_ids_cpu,
                 slot_mappings=slot_mappings,
             )
-            if isinstance(self.drafter, NgramProposer):
-                assert isinstance(sampled_token_ids, list), (
-                    "sampled_token_ids should be a python list when ngram is used."
-                )
-                draft_token_ids = self.drafter.propose(
-                    sampled_token_ids,
-                    self.input_batch.num_tokens_no_spec,
-                    self.input_batch.token_ids_cpu,
-                )
         elif spec_config.use_ngram_gpu():
             assert isinstance(self.drafter, NgramProposerGPU)
             (