AllDifferent to MILP: other formulations

[dourouc05]

Another formulation is https://www.andrew.cmu.edu/user/vanhoeve/papers/alldiff.pdf (section 6, page 35), without big-M; https://link.springer.com/article/10.1023/A:1014804110661 also has many formulations, with either a lot of variables or fewer.

However, this does not enlarge the set of features of this package, implementing other formulations may only improve the runtime performance. There should be a numerical study to give weights to the various bridges.