fix label handling during flatten

[CalCraven]

PR Summary: This code fixes a bug raised in #1206. This seems to be caused by the multiple hierarchies in this structure, which results in labels for each molecule. Then, after flattening, these compound labels refer to compounds not in the tree structure anymore, which get picked up after mb.clone is used on them, and causes errors in the particles themselves.

Need some feedback to suggest if just clearing out the labels is the correct approach for the flatten_hierarchy function. As you can see, that effects what you can access from the compound, but may be a limitation we're forced to accept.

PR Checklist

---

• Includes appropriate unit test(s)
• Appropriate docstring(s) are added/updated
• Code is (approximately) PEP8 compliant
• Issue(s) raised/addressed?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 85.69%. Comparing base (a4c41c5) to head (0019940).
Report is 8 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1208      +/-   ##
==========================================
- Coverage   85.73%   85.69%   -0.05%     
==========================================
  Files          53       53              
  Lines        4802     4795       -7     
==========================================
- Hits         4117     4109       -8     
- Misses        685      686       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[chrisjonesBSU]

Need some feedback to suggest if just clearing out the labels is the correct approach for the flatten_hierarchy function. As you can see, that effects what you can access from the compound, but may be a limitation we're forced to accept.

I think this is the correct approach, and is what I would expect after flattening the hierarchy.

[chrisjonesBSU]

LGTM! I tried this out with some similar applications discussed in the issue..I think the behavior now is what I would expect after calling flatten() on compound with children that had their own distinct labels.

[CalCraven]

Something sticky is happening with the labels on the Ubuntu tests. Will need to dig into this, as this PR needs to get merged to allow for the mBuild 1.0 GMSO PR to complete, which is holding up some GMSO PR's.