Text Normalizer Pipeline: make repeated text cleanup predictable

Normalization rules must be explicit. Choose Unicode NFC or NFKC deliberately, define locale-aware case conversion, standardize line endings and decide whether internal whitespace is preserved or collapsed. Deduplicate on the canonical form while retaining the first original value and its position for auditability. Case folding can merge strings that are distinct identifiers, so do not apply it blindly to keys, usernames or codes. Emit input, changed, duplicate and rejected counts, and enforce line and total-size limits.