Idea: sort found GT matches before constructing the (M)ILP problem

[maxkratz]

Since the pattern matcher GIPS uses runs highly parallelized, the found GT matches do not occur in a specific order. As a result, for multiple runs of the same GIPS problem, the GT matches can happen in different orders. This can result in differently generated (M)ILP problems, e.g., the order of the constraints can be different.

When the (M)ILP solver hits the configured timeout or finds multiple solutions with the same objective value, this can result in other solutions to the initial problem.

This issue proposes to "order" the GT matches before generating the (M)ILP problem to get a consistent result. As a downside, this may produce higher runtimes when a large quantity of matches is found by the pattern matcher.