Add max_input and max_tokens parameters to KnowledgeBrainQA constructor (#2247)

This pull request adds the `max_input` and `max_tokens` parameters to
the `KnowledgeBrainQA` constructor. These parameters allow for setting
the maximum input length and maximum number of tokens for the
KnowledgeBrainQA model. This provides more flexibility and control over
the input size and tokenization process.

backend/modules/brain/knowledge_brain_qa.py

@@ -181,6 +181,8 @@ class KnowledgeBrainQA(BaseModel, QAInterface):
             brain_id=brain_id,
             chat_id=chat_id,
             streaming=streaming,
+            max_input=self.max_input,
+            max_tokens=self.max_tokens,
             **kwargs,
         )