paper-dl turns PaSa exports into local PDFs

Hot take: this is not trying to be another paper search engine; it’s the glue that makes a good search result set actually useful on disk.

• –It fills the gap between PaSa’s ranked metadata export and a usable local corpus.
• –The workflow is simple: export JSON, run `paper-dl`, get a folder of PDFs plus `failed.txt` for anything that needs a second pass.
• –Concurrency and automatic skipping make it suitable for larger batches without babysitting.
• –Best fit is local research automation and retrieval pipelines, especially when you want PDFs rather than just links and abstracts.