Restore missing records and deduplicate corrected train split

Update train split with corrected merged dataset

Update dataset with corrected merged version

Credit NTSB CAROL as data source

Add cleaned 1043 records: stripped PDF footer artifacts, removed 2 duplicate/colliding-id records

Add README

initial commit