Resolve conflict with weight windows and global russian roulette

[GuySten]

Description

Currently the global neutron russian roulette (which is used when survival biasing is turned on) conflicts with the application of weight windows.
As suggested in #2773 (comment), this PR makes global neutron russian roulette apply only when neutrons are outside of the weight windows mesh boundaries.

Fixes #2773

Checklist

• I have performed a self-review of my own code
• I have run clang-format (version 15) on any C++ source files (if applicable)
• I have followed the style guidelines for Python source files (if applicable)
• I have made corresponding changes to the documentation (if applicable)
• I have added tests that prove my fix is effective or that my feature works (if applicable)

[pshriwise]

Thanks @GuySten! You are a machine. Some thoughts to consider here in the comments.

[pshriwise]

I think we may want to limit this to only the first condition. In the case that the weight window check points at collisions are disabled, I expect we'd see the same aggressive rouletting cited in #2773.

[GuySten]

The problem is that without that check we will not have global rouletting when we are outside the mesh boundary when collision checkpoints are disabled and surface checkpoints are enabled.

EDIT:
Maybe further restructure is needed.
I will think about that.

What do you think?

[pshriwise]

If I'm understanding right, with weight windows on we'll still check for a valid weight window bin in apply_weight_windows and in the absence of a valid bin (i.e. if the particle is outside the mesh).

It could be that particles exceed the lower or upper energy bounds of the weight windows rather than, but in essence I'd say this is equivalent to the argument about particles leaving the mesh -- if the user is interested in this energy space then it should be part of the weight windows definition.

Now, hypothetically, we could run into a problem where weight windows are enabled in the settings but all checkpoints are turned off, then no rouletting would apply. Maybe an extra initialization check/warning to address this would be a good idea.

[GuySten]

If I understand correctly, If weight windows are on and collision checkpoints are off we will have global rouletting when outside of mesh only in surface checkpoints which is weird behavior.
Usually we use global rouletting only in collisions.

[pshriwise]

It's different, true. Though the purpose of rouletting is that we address low-weight particles at some point in the simulation so they don't run for an exceedingly long amount of time, but I don't see a reason that this must occur during a collision event.

For infinite media problems we may never hit a surface to roulette these particles, but I also don't think we'd see application of weight windows in those cases.

That said, if you have a way of restructuring things so it aligns better with the current rouletting check points, I'm game! Main thing I'm watching out for is that we avoid an additional weight window look-up or additional particle state if possible, which your PR does well now.