[Simulator] Add high-fidelity CPU-based inference simulator (#33824)

Co-authored-by: zhouhaizhu.zhz <zhouhaizhu.zhz@alibaba-inc.com>
Co-authored-by: LinSiyuan814 <linsiyuan.lsy@alibaba-inc.com>
Co-authored-by: hzh0425 <hzh0425@apache.org>

tools/sglang-simulator/examples/workloads/sharegpt-example.json

@@ -0,0 +1,38 @@
+[
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "Explain prefix caching in one concise sentence."
+      },
+      {
+        "from": "gpt",
+        "value": "Prefix caching reuses KV states shared by prompt prefixes."
+      }
+    ]
+  },
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "What does time to first token measure?"
+      },
+      {
+        "from": "gpt",
+        "value": "It measures latency from request arrival to the first generated token."
+      }
+    ]
+  },
+  {
+    "conversations": [
+      {
+        "from": "human",
+        "value": "Why does decode latency matter for serving?"
+      },
+      {
+        "from": "gpt",
+        "value": "Decode latency determines the cadence at which later tokens reach the user."
+      }
+    ]
+  }
+]

tools/sglang-simulator/examples/workloads/timestamp-trace-example.jsonl

@@ -0,0 +1,3 @@
+{"prompt":[100,101,102,103],"prompt_len":4,"output_len":4,"timestamp":0}
+{"prompt":[100,101,102,104],"prompt_len":4,"output_len":4,"timestamp":75}
+{"prompt":[200,201,202,203],"prompt_len":4,"output_len":4,"timestamp":250}