Human Evaluation of Procedural Knowledge Graph Extraction from Text with Large Language Models