Implementing chromatic amplitude as an optimization term

[Björklund]

Hi,

I'll send one the next time the problem appears, it doesn't seem to be consistently reproducible.

To sum up the problem: the optimizer quits, and seems to think it has converged, sometimes when I optimize the chromatic derivatives of alpha and beta. When looking in the twiss output file, the output values of the derivatives say '-nan(ind)' when this happens.

Sorry for the incoherent description before.

//Jonas