Remove old train.parquet file and update new file with larger size; enhance convert.py to drop duplicates and reset index; add PTT article crawler script

Update train.parquet file and comment out CSV export in convert.py

Rename output files to 'train.csv' and 'train.parquet' for consistency