Monitor+Nonlinear invariants+Aperture yields SegFault

[cemitch99]

The test script below yields a SegFault. However, the script runs successfully in any of the following cases:

1. If the flag "nonlinear_lens_invariants=True" is changed to "nonlinear_lens_invariants=False".
2. If the Aperture element is removed or commented.
3. If the first Monitor element is removed or commented.

The following is a Python script. (The change .py -> .txt was made to allow attachment to this Issue.)
run_aperture_test_segfault.txt

[cemitch99]

This bug only affects users who model nonlinear integrable optics (NIO) with apertures.

[ax3l]

Ok, I see the issue. It's a bit intricate: for the nonlinear lens invariant calculation, because it is a per-particle property, we add extra runtime parameters to the beam (particle container). In the particle collection of lost particles we also deal with an extra runtime attribute, the position along s where the a particle was lost (s_lost). Both are not in sync, so there is some confusion that creates the crash. I'll clean that up.

[ax3l]

Developed a bug fix in #795.