Slow image extraction

[scriptcoded]

The current process for extracting images goes as follows:

1. Find images within the document
2. Render each page of the document as an image
3. Use sharp to extract the images from the rendered pages

This process is both slow and requires an external library (sharp), that in turn has a native dependency (libvips).

In an ideal world we would extract the images directly from the PDF. As of right now I've not found a way to do this, but perhaps we could get some clues from pdfcpu (https://github.com/pdfcpu/pdfcpu/blob/6b2e3b4ba26ed6839410ca2fd00f21cb4649efbe/pkg/pdfcpu/extract.go#L51).