Better meta-data recognition for pdfurl-jobs

[robamler]

Dontprint displays an icon in the address bar if the active browser tab is displaying a PDF file. Currently, Dontprint is not able to gather any meta-data (most importantly: journal name) for these kind of jobs. We may want to try to run the Zotero translators on history[-1] since this will in many cases be a journal website. (However, the actual PDF file should still be downloaded from the original URL, not from the URL extracted by the Zotero parser)