Move library collections as a group and map v2 identity refs.

Collapsed collection actions now move, rename, delete, and rerun every part so chains stay together. Identity mode follows the v2u3 Ref2VA picture wiring.

Co-authored-by: Cursor <cursoragent@cursor.com>

server/assets/workflow_qwen_edit.json

@@ -1,7 +1,7 @@
 {
   "1": {
     "inputs": {
-      "unet_name": "qwen_image_edit_2509_fp8_e4m3fn.safetensors",
+      "unet_name": "split_files\\diffusion_models\\qwen_image_edit_2509_fp8_e4m3fn.safetensors",
       "weight_dtype": "default"
     },
     "class_type": "UNETLoader",
@@ -39,21 +39,22 @@
     "class_type": "LoadImage",
     "_meta": { "title": "Load input still" }
   },
-  "5": {
+  "17": {
     "inputs": {
-      "text": "",
-      "clip": ["2", 0]
+      "prompt": "",
+      "clip": ["2", 0],
+      "vae": ["3", 0],
+      "image": ["4", 0]
     },
-    "class_type": "CLIPTextEncode",
-    "_meta": { "title": "Positive prompt" }
+    "class_type": "TextEncodeQwenImageEdit",
+    "_meta": { "title": "Qwen Image Edit prompt" }
   },
-  "6": {
+  "19": {
     "inputs": {
-      "text": "blurry, low quality, deformed, bad anatomy, watermark, text, extra fingers",
-      "clip": ["2", 0]
+      "conditioning": ["17", 0]
     },
-    "class_type": "CLIPTextEncode",
-    "_meta": { "title": "Negative prompt" }
+    "class_type": "ConditioningZeroOut",
+    "_meta": { "title": "Zero negative" }
   },
   "7": {
     "inputs": {
@@ -70,10 +71,10 @@
       "cfg": 1,
       "sampler_name": "euler",
       "scheduler": "simple",
-      "denoise": 0.75,
+      "denoise": 1,
       "model": ["11", 0],
-      "positive": ["5", 0],
-      "negative": ["6", 0],
+      "positive": ["17", 0],
+      "negative": ["19", 0],
       "latent_image": ["7", 0]
     },
     "class_type": "KSampler",