The import routine keeps it simple by only using the first entry when the book is being imported and the database entry is being created. Given that much of the other metadata needs to have that database entry before you can add them to the database, the second connect is pretty much required. Quite a bit of the data does not have Dublin Core metadata to even store it in a minimal .opf file. You can check
Dublin Core™ Metadata Element Set, Version 1.1: Reference Description for more information. Also note that not all DC metadata is required. For an ePub3 file, title, identifier, language and modified are required so in theory, you don't even need to have the creator name available.