Use the Link header to get a memento’s URL

[Mr0grog]

This fixes an issue where the Memento.url property could be slightly incorrect, since it was based on the URL you requested the memento from (e.g. https://web.archive.org/web/20221010000000/<url>), rather than the actual URL the memento was captured from. The URL the memento is requested from matches records via SURT key rather than the URL.

For example, requesting an archived copy of http://fws.gov/ might return a capture of https://www.fws.gov/ instead. The returned Memento object’s url property used to be http://fws.gov/ in this case, but this changes it to be https://www.fws.gov/.

Fixes #99.

[Mr0grog]

I think I might hold off on merging this for the weekend and consider adding a links property to Memento that is the parsed links, since I had to make use of them for this anyway. (See #57)

It also might make sense to parse the Memento-Datetime header (we can use Python’s built-in email.utils.parsedate_to_datetime function for that). OTOH, I don’t see any situation where I’d expect the timestamp we are already parsing from the Memento URL to be inaccurate like the capture URL was, so this may not really be meaningful.