Add HTTP retry mechanism and request timeout to transcribe_legacy

Use requests Session with Retry adapter (5 retries, backoff factor 1)
and add 30s timeout to both submit and query requests for better
reliability against transient network errors.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

transcribe_legacy.py

@@ -5,6 +5,8 @@ import time
 import uuid
 
 import requests
+from requests.adapters import HTTPAdapter
+from urllib3.util.retry import Retry
 from boto3 import client as s3_client
 from botocore.config import Config
 from dotenv import load_dotenv
@@ -22,6 +24,11 @@ S3_BUCKET = os.getenv("S3_BUCKET")
 SUBMIT_URL = "https://openspeech.bytedance.com/api/v3/auc/bigmodel/submit"
 QUERY_URL = "https://openspeech.bytedance.com/api/v3/auc/bigmodel/query"
 
+retry = Retry(total=5, backoff_factor=1, status_forcelist=[500, 502, 503, 504])
+session = requests.Session()
+session.mount("http://", HTTPAdapter(max_retries=retry))
+session.mount("https://", HTTPAdapter(max_retries=retry))
+
 s3 = s3_client(
     "s3",
     endpoint_url=S3_ENDPOINT,
@@ -116,7 +123,7 @@ body = {
     },
 }
 
-resp = requests.post(SUBMIT_URL, json=body, headers=headers)
+resp = session.post(SUBMIT_URL, json=body, headers=headers, timeout=30)
 logid = resp.headers.get("X-Tt-Logid", "")
 status = resp.headers.get("X-Api-Status-Code", "")
 print(f"Submit: {status}")
@@ -125,13 +132,14 @@ if status != "20000000":
     print(f"FAILED: {resp.headers.get('X-Api-Message', '')}")
 else:
     while True:
-        resp = requests.post(
+        resp = session.post(
             QUERY_URL,
             json={},
             headers={
                 **make_headers(task_id),
                 "X-Tt-Logid": logid,
             },
+            timeout=30,
         )
         code = resp.headers.get("X-Api-Status-Code", "")
         if code == "20000000":