Reconcile Records Across Sources

In an AI-assisted coding round, you are asked to reconcile records from multiple data sources. Each source contains person records with fields such as `name`, `email`, `phone`, and `address`. Records may have missing values, inconsistent formatting, duplicates, and conflicting information. Design and implement a reconciliation approach that: 1. decides whether two records refer to the same real-world person, 2. groups matching records, 3. produces a canonical merged record for each group, 4. leaves unmatched records separate. Be explicit about normalization, matching rules or confidence scoring, conflict resolution, and how you would validate the quality of the matches.