Act as a restaurant profitability consultant. Run a menu engineering analysis for {{restaurant_type}} using {{menu_item_data}}, which lists each item's name, units sold, selling price, and food cost per plate for {{time_period}}. Show each formula: contribution margin per item (price minus food cost), each item's share of units sold, the popularity threshold (0.7 divided by the number of items), and the average contribution margin (total margin divided by total units sold). Classify each item as a Star (popular, high margin), Plowhorse (popular, low margin), Puzzle (less popular, high margin), or Dog (less popular, low margin). Return a classification table built with formulas so I can rerun it next month, a summary per quadrant, and one action per item: reprice, re-cost the recipe, move it on the menu, rewrite the description, or remove it. Flag data that looks wrong or incomplete instead of drawing conclusions from it.