Ruin & recreate moves

[ge0ffrey]

PoC here: https://github.com/Christopher-Chianelli/timefold-solver/tree/feat/ruin-moves https://timefold.slack.com/archives/C04PQNRTJVD/p1720503735957709

We need to determine if these moves are useful and worth the cost.

• Does it improve the solutions for small problems, for large ones, or not at all?
• Is it better to ruin 10 %, 50 % or 90 % of the solution?
• How often should these run? Probabilistic selection to the rescue?

If/When we know that these moves work, we need to implement list variable support, and do the above all over again. If we save the data, this will be not only a nice new feature, but also a nice blog post.