fix: increase dlq buffer size

for our high throughput consumers, this is too small

it's important not to drop messages as we are going to
be using dlq more for stale messages too

fixes SENTRY-FOR-SENTRY-A1Z

arroyo/dlq.py

@@ -334,7 +334,7 @@ def __init__(
         self,
         policy: DlqPolicy[TStrategyPayload],
     ) -> None:
-        self.MAX_PENDING_FUTURES = 1000  # This is a per partition max
+        self.MAX_PENDING_FUTURES = 2000  # This is a per partition max
         self.__dlq_policy = policy
 
         self.__futures: MutableMapping[