Alignment of near miss 3 documentation/implementation with theory

[tsoumakas]

Hi, in the paper about near miss (https://www.site.uottawa.ca/~nat/Workshop2003/jzhang.pdf), a 4th variation is described, called "distant". The documentation of near miss 3, reads as if the original near miss 3 variation is merged with the 4th "distant" variation, perhaps because the writing of the original paper was not clear enough. The 4th variation was presented in the flow of near miss 3.

[glemaitre]

Indeed. The paragraph is super confusing. Since NearMiss-3 does not allow to have selection criteria, it is more of a cleaning method (in terms of class inheritance).

I think that we can deprecate version=3 and create a new class. Regarding the most distant strategy, we could also create a new class.

[chkoar]

I could take it over

[glemaitre]

Happy to review it then ;).