update neighbors to fix phagocytosis bug

[drbergman]

This is an alternate way to resolve #320. This is simpler than the way proposed in #338 but I think that one is a better rework.

Notes

• make sure every cell has an updated neighbors list to start
• this is the simplest change (fewest lines) to fix the bug

Warning

However, the interactions sample on Ubuntu has a 'std::bad_alloc' what(): std::bad_alloc` error thrown. I cannot reproduce this running on my Mac or a Linux-based HPC. I do not think this fix is triggering that error since none of the other tests are raising that error.

[rheiland]

I did not get an error running the ./interaction_demo on my Ubuntu (Ubuntu 24.04 LTS) with g++ 13.2.0.

[rheiland]

When I run the template project in this PR and the one in #338 on the .xml and cell_rules.csv that you shared in slack (with the usual 1 core, fixed seed), I see non-bitwise reproducible results: qualitatively "the same" for snapshot00000001.svg, but not identical. Did you see the same? If so, do you know why?

[drbergman]

Making sure we're on the same page: this is a comment about #338 rather than this one, right? In #338, I rearranged the ordering of update_all_cells so that the updating of the neighbors lists only needed to happen one time. I did this by moving the standard_cell_cell_interactions after the advance_bundled_phenotype (and after resolving the birth and death in there).

Though I guess I don't see how that would affect the template project since that should have no interactions...

[rheiland]

It was a comment about both #338 and #339 - comparing their sims on the model in question, expecting/hoping to see identical results, but only seeing "close". I think #338 may require a zoom walk-through (or else a good night's sleep for me). Seems like a lot of changes, but maybe some verbal explanation would help.

[drbergman]

closing in favor of # 338