CSV Clean-up + Data QA report (demo, sample data)

Demo on a fictional CRM contacts export (example.com emails, Ofcom drama-range phone numbers), not a client file. Written per-column rules clean the file: trimmed names, normalised emails, UK postcodes, E.164 phone numbers, ISO dates from explicit input formats, mapped plan names and money columns, plus de-duplication that keeps the most complete record. Outputs: cleaned, rejected (with a reason per row) and duplicates files, a cell-by-cell change log and a self-contained QA report. An independent verifier re-reads the outputs and confirms input rows = clean + rejected + duplicates (1,232 = 1,045 + 113 + 74). The same rules run unchanged on next month's export. Stack: TypeScript, Node, React, Vite.