What happens to metadata, bookmarks, and page size
A merge is not a “blend.” Each included page is copied from its source page dictionary — MediaBox, CropBox, rotation, content stream, resources. If PDF A is Letter (612×792 pt) and PDF B is A4 (595.28×841.89 pt), the merged file contains both boxes. A printer that “fits to page” will then scale one of them. That is why a merged packet can look fine on screen and print with a thin white strip or a clipped footer.
Document-level metadata (Author, Title, XMP) typically comes from the first file, or is omitted. Bookmarks and named destinations that pointed at “page 4 of file B” do not automatically reindex to “page 19 of the merge.” Links that used internal page numbers often go nowhere. If you need a table of contents, rebuild it after the merge.
Incremental updates in a source file (the leftover bytes after an old %%EOF) are not a feature we try to preserve. Copied pages come from the live page tree.
Orientation conflicts
A phone scan saved as a landscape MediaBox with /Rotate 90 will display upright in Chrome and come out sideways in some printers if the merge copies the dictionary as-is. Check the thumbnails here before you download. Exclude a blank scanner page while you are at it — those pages are often 2–4 MB of white JPEG each.