Fix attention mask handling in EoMT-DINOv3 converter

## Summary - update the conversion verifier to build an additive attention mask that mirrors the original EoMT implementation instead of multiplying by a boolean mask - seed PyTorch before running verification so repeated conversions are deterministic ## Testing - `python -m compileall src/transformers/models/eomt_dinov3/convert_eomt_dinov3_to_hf.py` ------ https://chatgpt.com/codex/tasks/task_b_68d65a86fc4c8336abf69a5c0d342a8a