Fix adq patch regression

[guilpier-code]

This PR intends to fix gopro ticket 1155.

Note that :

• Note a special branch branch-v8.8.0 was created so that this PR can be based on it. It seems that basing a PR on a tag is impossible.
• the fix was derived from the first commit where the regression occurs (which is somewhere in between v8.7.0 and v8.8.0). Apparently there are conflict with tag v8.8.0.
• This fix is a dirty quick hack : its purpose is to demonstrate the cause of an adequacy patch regression, introduced when removing a global variable. This PR mainly consists in restoring a call to function SIM_RenseignementProblemeHebdo.
• Carrying data required by this call turns out to be difficult, the resulting code is uglier than it was : optimization abstract method solve(...) has now more arguments that are not used in all derived implementation classes, specially the default class.
• If we decide to integrate these changes to develop branch, we need remove the previous identified flaws. That said, we may not need to apply this fix to develop : @JMJ-rte opinion is that adequacy patch isolation first step should be skipped, so we won't need a new call to SIM_RenseignementProblemeHebdo
• We validated this fix by running the study supplied in ticket 1155 while asking mps files printings, and comparing them (a use of md5sum is enough).
• This regression was not caught sooner because there seems to be no associated test. We should have a test validating this.