particles.at_element is off at turn 0 if tracking with ele_start option

[roman-martin]

It appears that the particles.at_element is not set correctly if the ele_start option of line.track is used.

As a consequence, to get element name of the lost particle, I need to reference

1. particles.at_element + start_element_index if lost on turn 0
2. particles.at_element if lost later

It would make more sense if particles.at_element would refer to the correct element right from the start.

I believe (re)initialization of particles.at_element should happen somewhere around here.