Always "embed", so JSON consumers can identify the command
Whether the run completed without errors
Whether this was a dry run, so no files were touched
Number of markdown files examined
Markdown files whose content was rewritten (empty in a dry run)
Number of image links rewritten to data URIs
Image files removed because nothing in the processed set references them (empty in a dry run)
Image files kept because a file in the processed set still references them
Error messages, one per failure
Warning messages
Summary of an embed run, in the shape emitted by
--json.