Rules/la generic

This PR presents a major rewrite of the description of rule la_generic: it corrects several typos, polishes the mathematical notation used (e.g., correct sub-indexes) and makes explicit some aspects of the algorithm that seemed to be implicit in the previous description.

Considering this PR, I also looked for the issues about la_generic that are still opened, to see if we can close them:

• @Lachnitt For this issue about the lack of clarity when explaining how to decide the resulting relation, after adding up every inequality: do you consider that we can close it?
• @Lachnitt From this issue, if I am not mistaken, the present version of la_generic could deal with your example linear inequalities. Should we close the issue?

@bandreotti @hjsc This issue is closed, but I am referring to it, since my description of how to choose the resulting relation is following Bruno's observation. Is this Ok?