requeue downloads rejected for platform concurrency instead of discarding them

v2.3.0 burned through all 498 models in minutes with download_failed=498: the
platform already had 8 downloads running, every create returned "用户最多同时运行
8 个下载任务", and the worker treated a create failure as terminal (CREATE_FAILED)
and moved on.

Now a concurrency rejection puts the model back at the head of the queue and
waits for the next poll, so only genuine per-model failures (uniqueness check,
etc.) are recorded as failed.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

main.py

@@ -115,6 +115,9 @@ TASK_PAGE_ENDPOINT = "/api/adapt/task/page"
 QUOTA_FULL_MSG     = "当前等待中或运行中的异步模型验证任务数量已达上限"
 _hygon_quota_full  = False   # zhoukaile 额度是否已打满
 _bi150_quota_full  = False   # bi150 兜底账号额度是否已打满
+# 平台下载并发上限提示（错误码 60005）；命中时把模型放回队列重试而不是丢弃
+DOWNLOAD_BUSY_MSG  = "个下载任务"
+_download_busy     = False
 GPU_TYPE_HYGON     = "hygon_k100-ai"
 GPU_TYPE_BI150     = "Iluvatar_bi-150"
 
@@ -2645,7 +2648,14 @@ def create_download_task(token: str, model_id: str) -> bool:
                 print(f"✅ 下载任务已提交: {model_id}", flush=True)
                 return True
             else:
-                print(f"⚠️ 下载任务业务失败 ({model_id}): {data.get('message')}", flush=True)
+                msg = str(data.get("message") or "")
+                if DOWNLOAD_BUSY_MSG in msg:
+                    # 平台侧下载并发已满，不是这个模型的问题，稍后重试
+                    global _download_busy
+                    _download_busy = True
+                    print(f"⏸️ 下载并发已满，稍后重试 ({model_id})", flush=True)
+                else:
+                    print(f"⚠️ 下载任务业务失败 ({model_id}): {msg}", flush=True)
                 return False
         else:
             print(f"❌ HTTP 错误 ({model_id}): {resp.status_code} - {resp.text}", flush=True)
@@ -2929,13 +2939,20 @@ def _download_phase(token, pending_models, active_models, completed_results) ->
                     _state["download_failed"] += 1
 
         # 2. 补充新任务（最多补到 MAX_CONCURRENT_DOWNLOADS 个）
+        global _download_busy
+        _download_busy = False
         while len(active_models) < MAX_CONCURRENT_DOWNLOADS and pending_models and not _shutdown.is_set():
             next_model = pending_models.pop(0)
             if create_download_task(token, next_model):
                 active_models.add(next_model)
                 print(f"▶️  启动下载: {next_model} (当前活跃: {len(active_models)})", flush=True)
+            elif _download_busy:
+                # 平台下载并发已满：放回队首等下一轮重试，不能当成失败丢弃，
+                # 否则整个列表会在几分钟内被空转消耗光（v2.3.0 曾因此 498 个全部“失败”）
+                pending_models.insert(0, next_model)
+                break
             else:
-                # 创建失败也视为完成（避免卡住）
+                # 模型本身的问题（如唯一性检查不通过），视为完成，避免卡住
                 completed_results[next_model] = "CREATE_FAILED"
                 _state["download_failed"] += 1
                 print(f"❌ 创建失败: {next_model}", flush=True)