low resolution simulated annealing

The code for the low resolution simulated annealing is kept together with a chewing-gum.

Known issues are:

• the difference between target and diagonal score (here, here, and here) is reverted; this cascade everywhere.
• because of the lack of a normalization factor, in the highlevel function (here) the target has been arbitrarily placed at the number of links present in the original matrix. This may be done better...