Implement Async Scheduling (#19970)

Signed-off-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>

vllm/v1/request.py

@@ -77,6 +77,7 @@ class Request:
         self.num_prompt_tokens = len(self.prompt_token_ids)
         self._output_token_ids: list[int] = []
         self._all_token_ids: list[int] = self.prompt_token_ids.copy()
+        self.num_output_placeholders = 0  # Used in async scheduling.
         self.spec_token_ids: list[int] = []
         self.num_computed_tokens = 0
         self.cache_salt: Optional[str] = cache_salt