docs: add contact information (email and WeChat)

Update citation with arXiv link and author list

Add pre-trained weights via Git LFS

Remove figures from README

Add framework, motivation, and visualization figures to README

Remove trainable parameter reduction claim from README highlights

Note AnomalyCLIP-compatible data format in README

Simplify Getting Started: merge train/test into single script entry

Rewrite README for CVPR 2026 camera-ready release

Simplify codebase and add README

Clean up: remove redundant args, logs, and dead code

Initial commit: VisualAD-CLIP anomaly detection framework