DocExtract: multiple fixes

• Fixes for --raw-references for refextract.

• Handle extra vars for taks_run_core.

• Adds missing import.

• Fixes arXiv records selection.

• Fixes storing last run date for arXiv records.

• Handles PoS(LAT2005)239.

• Removes B from volume for nucl.phys.proc.suppl.

• Adds math-ph to arXiv prefixes.

• Never re-extract curated records.

• Does not default to page 1 for special journals.

• When the page looks like a year (19xx or 20xx) then converts the reference to misc text.

• Remove citation splitting heuristics by authors.

• New way of splitting references.

• Extracts year from references.

• Creates a subfield "y" with the year when it is known.

• Reworks arXiv report numbers processing.

• Adds CMS report numbers.

• Adds CERN-2004-003 to recognized reports.

• Does not default to page 1 when looking for numerotation.

• Removes use of all and any (for Python 2.4 compatibility)

• Removes extra write_message.

• Handles Phys.Lett. 100B (1981), 117.

• Discards references with many lines.

• Handles refs with only a report number.

• Fixes [6] ATL-PHYS-INT-2009-110 reportnumber detection.

• Adds test for recognizing a journal/reportnumber/doi alone.

• Bump refextract version.

• Fixes references splitting.

• Fixes refs extraction from string.

• Handles refs that do not start at the beginning of a line.

• Increases allowed refs len to 9 lines.

• Fixes report numbers replacing.

• Fixes report number kb format.

• Increases max number of lines for a reference. (closes #966)

• Handles figures within refs.

• Adds RT tickets info to logs,

• Fixes extra vars passing to tasks.

• Updates stats message.

• Updates refextract tmp filename in inveniogc.

• Fixes -c option.

• Forces Inspire format for tests.

• Fixes api unit tests.

• Fixes for printing refs.

• Extract collaboration into $$c subfield if CFG_INSPIRE_SITE=1. (closes #958)

Conflicts:
modules/docextract/lib/docextract_task.py