Fix OpenAI parameter: use max_completion_tokens instead of max_tokens

Newer OpenAI models (including gpt-5-nano) require max_completion_tokens
parameter instead of the deprecated max_tokens parameter.

backend/app/services/openai_service.py

@@ -103,7 +103,7 @@ Remember: Provide helpful, accurate information to assist with operational quest
                 model=self.model,
                 messages=messages,
                 temperature=0.3,
-                max_tokens=1500
+                max_completion_tokens=1500
             )
 
             # Parse response
@@ -239,7 +239,7 @@ Remember: Provide helpful, accurate information to assist with operational quest
                 model=self.model,
                 messages=messages,
                 temperature=0.3,
-                max_tokens=1500,
+                max_completion_tokens=1500,
                 stream=True
             )