fix: use self.client.base_url in acompletion method

The acompletion method was incorrectly accessing self.base_url which
doesn't exist. Changed to self.client.base_url to match the sync
completion method.

Author: alt-glitch

rlm/clients/openai.py

@@ -84,7 +84,7 @@ async def acompletion(
             raise ValueError("Model name is required for OpenAI client.")
 
         extra_body = {}
-        if self.base_url == DEFAULT_PRIME_INTELLECT_BASE_URL:
+        if self.client.base_url == DEFAULT_PRIME_INTELLECT_BASE_URL:
             extra_body["usage"] = {"include": True}
 
         response = await self.async_client.chat.completions.create(