Fetch info that is stored in Refs for DocumentInfo

[saebekassebil]

This is a fix for #1447 - The PDFDoc.info now contains not the "raw" info dictionary, but a dictionary where every Ref has been fetched and therefore easier to work with. The raw dictionary can still be fetched from PDFDocModel.getDocumentInfo()

[andreasgal]

Shouldn't this be done lazily?

[saebekassebil]

We could do that, but that would mean that I'd have to implement a new DocumentInfo object - which is no problem, but I wasn't sure that we'd want that many more objects flying around?

I also thinks that this sounds more performance demanding that it really is - Most PDFs doesn't store DocumentInfo as Refs and even when they do, they rarely store more than 5-7 pieces of information.

[brendandahl]

This will be fixed by some changes I'm making for how dictionary and arrays handle xrefs. Instead of passing around the xref the dict/array will do the xref fetching. We've had several bugs related to this since people keep forgetting that nearly every dict entry or array entry can be an xref unless the spec explicitly says that the object must be all direct objects.

Hopefully I'll wrap this up today, if not early tomorrow morning.

[saebekassebil]

Great, thanks! Closing?

[brendandahl]

Closing because of #1488.