Twenty-one layouts, one per content shape
Every combination of content maps to one layout. Always.
Three lead modalities. Four count buckets. Three group modalities. Each combination maps to one layout, designed specifically for that content shape. Not a closest match. Not a sensible approximation. The layout built for exactly what you wrote.
Photograph, icon, or text. Determines the visual register of the section's opening.
One, two, three, or many. Determines the grid structure.
Photograph, icon, or text, resolved across the supporting items as a set.