- Graphs are in an ordered list
- Informations with start/end and number of nodes
- User can choose to compute with conditional or distributional distance

[FIX] BUG ISI Parser (which depends on RIS parser)
      Adding a space in the regex breaks ISI parser
      Then the question is: is the space really usefull for RIS parser ?

unittests: revert the use of setUpClass for test node creation: the idea was valid but it triggers bug as in code.djangoproject.com/ticket/23325

unittests: mini_test_samples for parsers: pubmed, europresse, istex and ris (zotero variant) working now with tests_090_toolchain

risparser: correct ris tags comparing to specification in 'https://en.wikipedia.org/wiki/RIS_(file_format)'

taggers: french/english melt variants transformed from custom constructor into custom subclasses (in order for them to be compatible with constants.load_tagger)