v0.2.0

Added

• First public release of LoRA-EffSAM: parameter-efficient brain-tumor MRI segmentation with a
  LoRA-adapted EfficientSAM encoder, hybrid CNN-ViT decoder, size-adaptive boundary-aware loss,
  MC-Dropout uncertainty and explainability analysis.
• lora-effsam CLI with stages for data preparation, training, evaluation, statistics,
  uncertainty analysis and explainability.
• Scripts to fetch EfficientSAM and the Figshare dataset, run the full pipeline and reproduce
  every table, statistic and figure in results/.
• Experiment records, tables and figures for all reported runs in results/.