Tf/optimization

[tfarago]

Optimization is now a little bit more rich on algorithms.

[matze]

This is the only thing that bothers me with the changes. The maximizer already knows that position has to be optimized, hence it should be possible that the maximizer passes this information to the actual algorithm. For me it looks a bit redundant to pass the Parameter object twice.

If we could improve this, I would'n object to merge this into master.

[tfarago]

True, the halver was too simple and required a real x position at a time, because of the fact the f(x) domain is not infinite thanks to limiters. I improved the algorithm, so now it follows the general optimization algorithm signature: alg(x_0, *args, **kwargs), where x_0 is a value, not a parameter. This way our optimizers stay fairly general.