refactor: optimize scheduler policy lookup with try/except pattern

Replace double dictionary lookup pattern with more efficient try/except
approach. This eliminates redundant key existence check and improves
performance by reducing dictionary access from two operations to one
in the success path.

Changes:
- Use try/except KeyError instead of 'in' check followed by access
- Add 'from None' to suppress exception chaining for cleaner error messages
- Maintain identical error handling behavior with ValueError for unknown policies

Signed-off-by: zitian.zhao <[email protected]>

Author: skyloevil

vllm/v1/core/sched/scheduler.py

@@ -106,9 +106,10 @@ def __init__(
         self.requests: dict[str, Request] = {}
         # Scheduling policy
         policy_name = self.scheduler_config.policy
-        if policy_name not in self._POLICY_MAPPING:
-            raise ValueError(f"Unknown scheduling policy: {policy_name}")
-        self.policy = self._POLICY_MAPPING[policy_name]
+        try:
+            self.policy = self._POLICY_MAPPING[policy_name]
+        except KeyError:
+            raise ValueError(f"Unknown scheduling policy: {policy_name}") from None
         # Priority queues for requests.
         self.waiting = create_request_queue(self.policy)
         self.running: list[Request] = []