[Feature] Sglang Tracing: Fine-Grained Tracking for Request Latency - Part 1 (#9962)

Signed-off-by: Feng Su <sufeng@linux.alibaba.com>
Signed-off-by: Huaixin Chang <changhuaixin@linux.alibaba.com>
Signed-off-by: Peng Wang <rocking@linux.alibaba.com>

python/sglang/srt/managers/io_struct.py

@@ -605,6 +605,9 @@ class TokenizedGenerateReqInput:
     # Image gen grpc migration
     return_bytes: bool = False
 
+    # tracing context
+    trace_context: Optional[Dict] = None
+
 
 @dataclass
 class BatchTokenizedGenerateReqInput:
@@ -654,6 +657,9 @@ class EmbeddingReqInput:
     # For background responses (OpenAI responses API)
     background: bool = False
 
+    # tracing context
+    trace_context: Optional[Dict] = None
+
     def normalize_batch_and_arguments(self):
         # at least one of text, input_ids, or image should be provided
         if self.text is None and self.input_ids is None and self.image_data is None: