feat(attack-ui): add ground-truth aware evaluation and sample GT auto-selection

feat(metrics): add rollout-based attention shift metrics (W1 and JSD)

feat(apuana): add local-only cluster tab flow, history diagnostics, and metric selection updates

fix: random_start=False em PGD para reprodutibilidade

git commit -m "feat: expand ViTViz with experiment pipeline, metrics module, model loading improvements and report UX" -m "- add experiment infrastructure with configs, sweep scripts, notebooks and Makefile tasks

feat: make possible to use different ViT formats and architectures

feat: adds different models capability

feat: new version of report after attack

feat: adds hosting capabilities

feat: implements TGR attack with settable parameters

feat: adds support to resnet gradients in SAGA attack

fix: fixes mim attack, putting attention capture inside the attack loop

feat: adds visualization by head/layer

fix: saga attack is now correct

chore: organzing SAGA attack

fix: fixes MIM attack to init momentum on denormalized space

feat: adds MIM attack

feat: add new SAGA attack with attention capture

feat: add fgsm attack and model conversion file from timm to pytorch

feat: primeira versão