feat: 反馈加来源/场景/运营回复 + 提现类型筛选 (#105)

用户反馈后台:区分反馈类型(比价反馈/普通反馈)+ 审核可给用户留言;提现后台:按提现类型筛选。

- feedback 表加 source(profile/comparison)/scene/admin_reply + 迁移;提交接口 /api/v1/feedback
  接收 source/scene,来源判定显式 source 优先、否则据 scene 有无派生(有=comparison);
  /records 带回 scene + admin_reply。
- admin 反馈:列表加「反馈类型」筛选;采纳/拒绝支持存 admin_reply(给用户的回复,用户端可见);
  FeedbackOut 带出 source/scene/admin_reply。
- admin 提现:WithdrawOut 暴露 source,列表加「提现类型」筛选(coin_cash=福利页提现 / invite_cash=邀请提现)。
- 补 4 项测试:来源派生、回复存取、反馈按类型筛选、提现按类型筛选。

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: zzhyyyyy <2685922758@qq.com>
Reviewed-on: #105
Co-authored-by: zhuzihao <zhuzihao@wonderable.ai>
Co-committed-by: zhuzihao <zhuzihao@wonderable.ai>

tests/test_admin_write.py

@@ -368,6 +368,35 @@ def test_reject_feedback_records_reason_and_no_coin(
         db.close()
 
 
+def test_feedback_review_stores_admin_reply(
+    admin_client: TestClient, operator_token: str
+) -> None:
+    """审核可给用户留言(admin_reply,用户端可见),采纳/拒绝都能带,并回显在响应里。"""
+    fid = _seed_feedback("13900000021")
+    r = admin_client.post(
+        f"/admin/api/feedbacks/{fid}/approve",
+        json={"reward_coins": 100, "note": "内部备注", "reply": "感谢反馈,已采纳~"},
+        headers=_auth(operator_token),
+    )
+    assert r.status_code == 200, r.text
+    assert r.json()["admin_reply"] == "感谢反馈,已采纳~"
+    db = SessionLocal()
+    try:
+        fb = db.get(Feedback, fid)
+        assert fb.admin_reply == "感谢反馈,已采纳~" and fb.review_note == "内部备注"
+    finally:
+        db.close()
+
+    fid2 = _seed_feedback("13900000022")
+    r = admin_client.post(
+        f"/admin/api/feedbacks/{fid2}/reject",
+        json={"reason": "无法复现", "reply": "已收到,后续跟进"},
+        headers=_auth(operator_token),
+    )
+    assert r.status_code == 200, r.text
+    assert r.json()["admin_reply"] == "已收到,后续跟进"
+
+
 # ===== admin 账号管理(super_admin) =====
 
 def test_create_and_update_admin(admin_client: TestClient, super_token: str) -> None: