[Model][Speculative Decoding] DeepSeek MTP spec decode (#12755)

Signed-off-by: Lu Fang <fanglu@fb.com>
Co-authored-by: LiuXiaoxuanPKU <lilyliupku@gmail.com>

vllm/sequence.py

@@ -1307,6 +1307,8 @@ class ExecuteModelRequest(
     previous_hidden_states: Optional[HiddenStates] = None
     # The number of forward steps to run.
     num_steps: int = 1
+    # The step index for spec model input.
+    spec_step_idx: Optional[int] = None
     # Finished request ids since last step.
     finished_requests_ids: List[str] = msgspec.field(default_factory=list)
     # The last sampled token ids for multi step decoding.