fix(dispatcher): re-raise ZeroGPUQuotaError caught by generic except handlers

Quota stops raised inside the Gradio SSE loop and the async-jobs poll loop
were swallowed by their own 'except Exception' handlers, downgraded to
None and finally to a generic RuntimeError. The cron therefore marked all
6 tracks 'failed' on 09-12/09-13 instead of saving resume state and
deferring to the next day, burning zero of the remaining quota.

Re-raise ZeroGPUQuotaError explicitly in both loops so the nightly cron
persists resume state and recovers the album automatically.

moldovan-qwen/data/studio_settings.json

@@ -1,125 +0,0 @@
-{
-  "hf_token": "",
-  "openrouter_key": "",
-  "minimax_key": "",
-  "elevenlabs_key": "",
-  "fal_key": "",
-  "active_models": {
-    "chat": "Qwen/Qwen2.5-7B-Instruct",
-    "music": "MiniMaxAI/MiniMax-Music3",
-    "tts": "edge-tts-ro-RO",
-    "video": "Wan-AI/Wan2.1-T2V-14B",
-    "image": "black-forest-labs/FLUX.1-schnell"
-  },
-  "providers_enabled": {
-    "huggingface": true,
-    "openrouter": true,
-    "minimax": false,
-    "elevenlabs": false,
-    "local_fallback": true
-  },
-  "custom_endpoints": {
-    "whisper": "http://homelab:8086/v1/transcribe",
-    "flow_space": "https://abalanescu-flow2.hf.space/v1"
-  },
-  "custom_providers": [
-    {
-      "name": "Local vLLM",
-      "base_url": "http://100.70.158.21:8000/v1"
-    }
-  ],
-  "model_catalog": {
-    "audio": [
-      {
-        "id": "MiniMaxAI/MiniMax-Music3",
-        "label": "MiniMax Music 3 (ZeroGPU Neural Song)",
-        "provider": "huggingface",
-        "capabilities": [
-          "music"
-        ]
-      },
-      {
-        "id": "edge-tts-ro-RO",
-        "label": "Edge-TTS ro-RO (Emil)",
-        "provider": "edge",
-        "capabilities": [
-          "tts"
-        ]
-      }
-    ],
-    "video": [
-      {
-        "id": "Wan-AI/Wan2.1-T2V-14B",
-        "label": "Wan 2.1 Text to Video",
-        "provider": "huggingface",
-        "capabilities": [
-          "video"
-        ]
-      },
-      {
-        "id": "Wan-AI/Wan2.1-T2V-1.3B",
-        "label": "Wan 2.1 Fast Video",
-        "provider": "huggingface",
-        "capabilities": [
-          "video"
-        ]
-      },
-      {
-        "id": "THUDM/CogVideoX-5b",
-        "label": "CogVideoX 5B",
-        "provider": "huggingface",
-        "capabilities": [
-          "video"
-        ]
-      },
-      {
-        "id": "Lightricks/LTX-Video",
-        "label": "LTX Video",
-        "provider": "huggingface",
-        "capabilities": [
-          "video"
-        ]
-      },
-      {
-        "id": "black-forest-labs/FLUX.1-dev",
-        "label": "FLUX.1 Dev",
-        "provider": "huggingface",
-        "capabilities": [
-          "image",
-          "keyframes"
-        ]
-      },
-      {
-        "id": "black-forest-labs/FLUX.1-schnell",
-        "label": "FLUX.1 Schnell",
-        "provider": "huggingface",
-        "capabilities": [
-          "image",
-          "keyframes"
-        ]
-      }
-    ],
-    "text": [
-      {
-        "id": "Qwen/Qwen2.5-72B-Instruct",
-        "label": "Qwen 2.5 72B",
-        "provider": "huggingface",
-        "capabilities": [
-          "chat",
-          "lyrics",
-          "scripts"
-        ]
-      },
-      {
-        "id": "Qwen/Qwen3-32B",
-        "label": "Qwen3 32B",
-        "provider": "huggingface",
-        "capabilities": [
-          "chat",
-          "lyrics",
-          "scripts"
-        ]
-      }
-    ]
-  }
-}

moldovan-qwen/engine/model_dispatcher.py

@@ -606,6 +606,8 @@ class ModelDispatcher:
                                 if dl_resp.status_code == 200 and len(dl_resp.content) > 1000:
                                     return dl_resp.content
                         return None
+        except ZeroGPUQuotaError:
+            raise  # account-level quota stop: must reach the cron, never swallow
         except Exception as e:
             print(f"[ModelDispatcher] Gradio queue streaming error: {e}")
             return None
@@ -685,6 +687,8 @@ class ModelDispatcher:
                         if 'exceeded your Pro ZeroGPU quota' in _fail_text or 'reached its GPU limit' in _fail_text:
                             raise ZeroGPUQuotaError(_fail_text[:300])
                         return None
+            except ZeroGPUQuotaError:
+                raise  # account-level quota stop: must reach the cron
             except Exception as e:
                 print(f"[ModelDispatcher] Async job poll warning: {e}")
         return None