Evaluate OpenAI Decisions on all three public image attack sets

Complete image perturbation coverage on sixteen decision models

Add seven decision-model evaluations on the frozen curated release

Show clean and attacked images first in Dataset Viewer

Prepare public release and note forthcoming paper

Release v1.2.0: add 13 human-reviewed image coverage supplements

v1.1.0: remove overperturbed Common images consistently across scoring items

Simplify homepage quick start and move usage details to documentation

Release five curated DecisionRobust configurations v1.0.0