Fix: Use official HF Spaces preload_from_hub for model caching

Root cause:
- Dockerfile manual download didn't specify cache_dir parameter
- Models downloaded to wrong location (not /home/user/.cache/huggingface/hub)
- local_files_only=True couldn't find files → runtime failure

Solution (Following HF Spaces official documentation):
- Use preload_from_hub feature in README.md (official recommended way)
- Remove manual download code from Dockerfile
- preload_from_hub automatically downloads to ~/.cache/huggingface/hub
- config.py uses HF_HUB_CACHE which matches this path perfectly

Official documentation:
"The preload_from_hub feature allows you to pre-download models during build"
Source: https://huggingface.co/docs/hub/spaces-config-reference

Changes:
- README.md: Added preload_from_hub for both model repos
- Dockerfile: Removed lines 100-119 (manual download code)
- Dockerfile: Added comment explaining preload_from_hub usage

Expected behavior:
✅ Models preloaded during build to ~/.cache/huggingface/hub
✅ Runtime finds models via HF_HUB_CACHE environment variable
✅ CLIP model loads successfully without internet access

Updated: Dockerfile

Dockerfile

@@ -97,26 +97,8 @@ RUN ls -la $HOME/app/code/ImportLDraw/__init__.py || \
 RUN pip install --no-cache-dir --upgrade pip && \
     pip install --no-cache-dir -r requirements.txt
 
-# 预下载模型到默认缓存目录（构建时一次性下载，避免运行时卡顿）
-RUN echo "📥 Pre-downloading models during build (this may take 2-3 minutes)..." && \
-    python3 -c "\
-from huggingface_hub import hf_hub_download; \
-from transformers import CLIPModel, CLIPProcessor; \
-print('Downloading shape_gpt.safetensors (7.17 GB)...'); \
-hf_hub_download(repo_id='0xZohar/object-assembler-models', filename='shape_gpt.safetensors'); \
-print('✓ shape_gpt downloaded'); \
-print('Downloading shape_tokenizer.safetensors (1.09 GB)...'); \
-hf_hub_download(repo_id='0xZohar/object-assembler-models', filename='shape_tokenizer.safetensors'); \
-print('✓ shape_tokenizer downloaded'); \
-print('Downloading fine-tuned adapter (1.68 GB)...'); \
-hf_hub_download(repo_id='0xZohar/object-assembler-models', filename='save_shape_cars_whole_p_rot_scratch_4mask_randp.safetensors'); \
-print('✓ adapter downloaded'); \
-print('Downloading CLIP model (~600 MB)...'); \
-CLIPModel.from_pretrained('openai/clip-vit-base-patch32', use_safetensors=True); \
-CLIPProcessor.from_pretrained('openai/clip-vit-base-patch32', use_safetensors=True); \
-print('✓ CLIP downloaded'); \
-print('✅ All models pre-downloaded to ~/.cache/huggingface')" && \
-    echo "✅ Model weights cached successfully"
+# 注意：模型通过 README.md 的 preload_from_hub 自动下载（HF Spaces 官方推荐方式）
+# 下载到 ~/.cache/huggingface/hub/ 并在运行时通过 HF_HUB_CACHE 环境变量访问
 
 # 创建启动脚本（先启动 Xvfb，再运行应用）
 RUN echo '#!/bin/bash\n\