Group contacts by a normalized name key — lower-cased, with whitespace runs collapsed and ends trimmed — so "John Smith", "John Smith", and "JOHN SMITH" are one person. Return norm_name, n (how many contacts), and ids (comma-joined contact_ids ascending). Sort by norm_name.