Fix for issues #268 and #257 and parallel test for nodal swarm particle loss

[jcgraciosa]

Added logic and wrapper functions for handling the return of swarm DM getField() to handle changes from PETSc >= 3.22.0.

Also added:

• a wrapper function for swarm DM restoreField for symmetry with the getField function
• parallel test to check if there is nodal swarm particle loss when particles are advected beyond the neighbouring processor

[knepley]

@lmoresi @jcgraciosa I finally got this fixed (it was blocked by another MR): https://gitlab.com/petsc/petsc/-/merge_requests/8024

[lmoresi]

@knepley - is there any released version of petsc4py that has the incompatible return value ? We discussed whether we needed to exclude those versions or have this wrapper until they age out.

[knepley]

@lmoresi I do not believe it is. I checked 3.22.1, 3.22.0, and 3.21.6. It was not in any of those, so you should be fine to ignore it.

[knepley]

It might be good to setup an Underworld build from main to make sure I don't break anything.

[lmoresi]

@knepley — are you talking about having that as part of the PETSc testing, or as part of uw. If the latter, just on a daily cycle, or triggered by your changes ?

[knepley]

@lmoresi I was thinking one of the Underworld builds could target PETSc main. Probably once a week is fine. We could try to trigger it on specific changes, but I think we want to catch everything.