Weight Windows using in DAGMC results in lost particles

[shimwell]

Bug Description

When making weight windows with the nice new FW-CADIS and Random Ray method then using those weight windows in a model that has a source term inside a DAGMC universe I get lost particles.

After investigating with much help from @jtramm (many thanks for all the help) we tracked the bug down to this PR. When compiling on the current develop branch with the PR merged in I get lost particles, when commenting out the code and compiling again the model runs nicely.

While investigating I also noticed that we don't currently have a test which runs weight windows generated with FW-Cadis and Random Ray with DAGMC geometry. Most of the testing is in the regressions test folder. I would be tempted to take the minimal example linked below and add it to this test folder if that is an acceptable unit test?

Steps to Reproduce

Reproduce by running this minimal example with the dagmc in the same folder

Environment

ubuntu 25.04, openmc develop, python 3.13

[pshriwise]

I cloned the example and gave it a run, but I didn't run into any issues during random ray transport with the latest version of develop in OpenMC (commit bca818c at the time of this post).

Maybe I've got something configured differently?

[shimwell]

Thanks for taking a look, I shall double check everything over here. I was pretty sure it was failing with the latest develop commit. I updated the test PR #3504 which will also check for us