Add Image Anomaly Detetction benchmarks

It would be nice to add some benchmarks for the (Image) Anomaly Detection task. The CVPR 2026 VAND 4.0 Challenge has just concluded (https://benchmark.mvtec.com/vand2026-leaderboard) and it would be interesting to see how the winning approaches compare to the current sota models like PaDiM, PatchCore, ReConPatch, AnomlayDINO, etc. It would also be interesting to see how feature-based models compare to reconstruction-based models. Thanks in advance!