Unsuccessful VMEC calculation leads to segmentation fault in ONE_ITER optimization

[zhucaoxiang]

There is a known issue. One can use STELLOPT to do an one_iter optimization, like following the process VMEC -> BOOZ_XFORM -> NEO. If VMEC fails and because it is one_iter, STELLOPT will call stellopt_clean_up first in stellopt_fcn before calling stellopt_neo. Then a segmentation error will happen because we are trying to access eff_ripple before it was allocated in stellopt_neo. The flow chart is stellopt_clean_up -> stellopt_load_targets -> chisq_neo.

We should add some lines to catch this or stop this. Any suggestions? @lazersos

[tmqian]

I think this issue is specific to ONE_ITER (as Caoxiang mentioned).
Running the same file in LMDIF gives "first run fails, improve input" and no segfault.

So there might be a ready-made catch that can be copied from the LMDIF routine?

[zhucaoxiang]

@tmqian You are running my fixed version on Eddy. I will try to push it back to the main branch.