Your commands to process PDF with Ghostscript are lossy (they lose lots of metadata and in minor ways they also change how the PDF renders), and they produce very large PDF files.