Edit: to allow my question to be more useful to future researchers with similar questions, I've split my long, multi-paragraph series of questions into a pair of shorter questions. These posts can be found at the following links:
Why is predicted best k for de novo assembly so different from actual best k?
Large discrepancy between de novo assembly versus actual biological genome size
Why is predicted best k for de novo assembly so different from actual best k?
Large discrepancy between de novo assembly versus actual biological genome size