refextract: attempt at integrating giva

• Attemp at integrating Giva functionality into refextract.
• Improvements made to '--authors' mode, but no LaTeX support yet.
• Uses affiliations to support the identification of ambiguous authors.
• Work still needed to improve the reliability of the seeking of the end of author section (uses likely ending keywords, such as 'Abstract')
• Added '--affiliations' mode, which will try to extract affiliations from

a document

• Moved the authextract-specific tests into the new 'refextract_authextract_tests.py' file, since it the name was clashing with the current refextract_tests.py test suite.
• Refextract-config holds some of the institution names which are used to find affiliations, and were also held inside Giva.