Configurable max_tokens/max_completion_tokens key (#399)

<!--
Include a short paragraph of the changes introduced in this PR.
If this PR requires additional context or rationale, explain why
the changes are necessary.
-->

Makes the `max_tokens` request key configurable through an environment
variable per endpoint type. Defaults to `max_tokens` for legacy
`completions` and `max_completion_tokens` for `chat/completions`

<!--
Provide a detailed list of all changes introduced in this pull request.
-->
- Add the `GUIDELLM__OPENAI__MAX_OUTPUT_KEY` config option which is a
dict mapping from route name -> output tokens key. Default is
`{"text_completions": "max_tokens", "chat_completions":
"max_completion_tokens"}`

<!--
List the steps needed to test this PR.
-->
-

<!--
Link any relevant issues that this PR addresses.
-->
- Closes #395
- Closes #269
- Related #210

---

- [x] "I certify that all code in this PR is my own, except as noted
below."

- [ ] Includes AI-assisted code completion
- [ ] Includes code generated by an AI application
- [ ] Includes AI-generated tests (NOTE: AI written tests should have a
docstring that includes `## WRITTEN BY AI ##`)

---------

Signed-off-by: Tyler Michael Smith <[email protected]>
Signed-off-by: Samuel Monson <[email protected]>
Co-authored-by: Tyler Michael Smith <[email protected]>

Author: sjmonson

src/guidellm/backends/openai.py

@@ -30,6 +30,7 @@
     GenerationResponse,
 )
 from guidellm.scheduler import HistoryT, ScheduledRequestInfo
+from guidellm.settings import settings
 
 __all__ = ["OpenAIHTTPBackend", "UsageStats"]
 
@@ -628,12 +629,10 @@ def _get_body(
         # Handle token limits
         max_tokens = max_output_tokens or self.max_output_tokens
         if max_tokens is not None:
-            body.update(
-                {
-                    "max_tokens": max_tokens,
-                    "max_completion_tokens": max_tokens,
-                }
+            max_output_key = settings.openai.max_output_key.get(
+                endpoint_type, "max_tokens"
             )
+            body[max_output_key] = max_output_tokens
             # Set stop conditions only for request-level limits
             if max_output_tokens:
                 body.update({"stop": None, "ignore_eos": True})

src/guidellm/settings.py

@@ -89,6 +89,10 @@ class OpenAISettings(BaseModel):
     base_url: str = "http://localhost:8000"
     max_output_tokens: int = 16384
     verify: bool = True
+    max_output_key: dict[str, str] = {
+        "text_completions": "max_tokens",
+        "chat_completions": "max_completion_tokens",
+    }
 
 
 class ReportGenerationSettings(BaseModel):