* fix: Also extract document names from XML seriesInfo attributes
The old code only looked in the anchor string for the document names of
references, which doesn't work if the anchor uses a mnemonic. This caused lots
of missed references for many documents.
* No need to import lxml anymore
* Add tests
* Handle xinclude to bibxml URLs
* Wrap line
* Apply suggestion from @rjsparks
* Undo erroneous additions
* Address suggestion from @rjsparks
Look at v2 'title' attribute in reference type heuristics for XML drafts. Related to #3529.
- Legacy-Id: 19897
Note: SVN reference [19895] has been migrated to Git commit ea79fe0dcc183bc5cd8b27da67865c300b9dce4e
