Everything in Its Place: Benchmarking Spatial Intelligence of Text-to-Image Models | KnowAI Space