Fix issues with MeasuredRequestTimings model validation

Author: sjmonson

src/guidellm/backend/objects.py

@@ -140,7 +140,7 @@ def preferred_output_tokens(
             return self.response_output_tokens or self.request_output_tokens
 
 
-@MeasuredRequestTimings.register()
+@MeasuredRequestTimings.register("generation_request_timings")
 class GenerationRequestTimings(MeasuredRequestTimings):
     """Timing model for tracking generation request lifecycle events."""
 

src/guidellm/scheduler/objects.py

@@ -121,8 +121,8 @@ def __pydantic_schema_base_type__(cls) -> type[MeasuredRequestTimings]:
 
     schema_discriminator: ClassVar[str] = "timings_type"
 
-    timings_type: ClassVar[Literal["measured_request_timings"]] = (
-        "measured_request_timings"
+    timings_type: Literal["measured_request_timings"] = Field(
+        description="Type identifier for the timing measurement",
     )
     request_start: float | None = Field(
         default=None, description="When the backend began processing the request"