change: paper results win on conflict (official results immutable)

feat: merge real (paper) results with new community results

fix: results mapping returns [] when a run produced no results

feat: run the eval worker inside the Space (CPU, isolated venv)

feat: complete the eval engine — real worker + OEB-aware queue reader

fix: treat a missing/empty eval queue as empty, not an error

fix: self-bootstrap the private eval queue dataset on submit

feat: OpenEEGBench submission flow (arena runs the eval)

Fill model metadata in sample_results, drop unused fields

Fix TUEV category: event classification, not pathology detection

Align internal dataset IDs with HF dataset IDs

Align benchmark display names with paper

Align benchmark registry with results JSON

Add results from paper

Make it clear that the results are fake (#3)

Fix dataset IDs to match actual HuggingFace repo names

Add HF eval-results benchmark registry and API endpoint

Fix server 500 error and replace HF logos with eegdash branding

Align frontend and backend with adapter_finetuning paper

Migrate from Gradio to React + FastAPI architecture