Energy loss tracking

[swhite2401]

I has been brought to my attention that in some instances particles do not make it through one turn if cavities are off.
In this case at.ELossMethod.TRACKING cannot be used.
it is presently the default in find_orbit() and is therefore a serious issue.

This PR solves this issue by calculating the energy/element in case the particle is lost in less than a turn. This algorithm can be refined by smart iterative splitting of the ring based on loss index.

This also raises the question whether TRACKING should be the default, especially in find_orbit() for which it is used only to determine the initial guess: I would personally suggest to use INTEGRAL but this can be discussed.

@lfarv what do you think? Any better ideas?

[lfarv]

@swhite2401 : the problem with the INTEGRAL method in find_orbit6 is that it will always take all the dipoles/wigglers into account, while you can call find_orbit6 without dipole radiation (non-radiating machine), or with only some of them (why?). The TRACKING method will consider exactly what the user selects. For orbit guess, TRACKING is safer because it always matches the conditions for which the orbit is searched.

One could condition the 'integral' contribution of each dipole with its passmethod, but it would break the possibility of computing the energy loss independently of the radiation state of the lattice… It's nice to have all the lattice parameters available whatever the radiation state.

So given your patch, I would leave the defaults as they are. For marginal cases like the one you are mentioning, the user can still change the method: that's why the choice is proposed. Maybe a clear error message if the tracking fails would be helpful.

[swhite2401]

So, this one is ok for merging?

[lfarv]

@swhite2401 : yes, ok for me.

[swhite2401]

Ok I merge