Kicked out pkl upload, fixed csv-upload. Update requirements.txt to specify torch version and add gradio dependency.

Enhance data processing in app.py and openalex_utils.py by improving handling of referenced works and filling missing publication values with spaces.

Enhance data processing by adding CSV and Pickle file upload support, improving error handling, and refining the prediction pipeline. Update the `process_records_to_df` function to handle existing DataFrames and ensure required fields are processed correctly.

drop duplicates

Temorary change from medioid to mean

first commit