Refactor model references and enhance ZeroGPU support; update README and requirements for HF

Enhance RSA data handling: parse listener/speaker distributions from CSV, integrate into review metadata, and update visualization logic for improved agreement representation.

Enhance paper title handling: load titles from raw data CSVs for improved metadata accuracy in reviews

Update model loading to prioritize local trained models for improved accuracy; adjust configuration for model paths and fallback options

Enhance rebuttal handling in preprocessed_scores: track review numbers and improve loading logic

Add support for rebuttals and section header exclusion in review processing

Fix model loading utility and improve error handling in scoring pipeline

Move all pipeline scripts into pipeline/ directory for cleaner project structure