Fix inverse neighborhood comm

[MarcelKoch]

This PR fixes the generation of the inverse of a neighborhood communicator. Instead of the one-sided communication, it uses an all-to-all communication.

Some MPI implementations seem to take issues with the one-sided implementation.

[pratikvn]

Which MPI implementations have issues?

[MarcelKoch]

@yhmtsai noticed it. Perhaps it's also based on UCX.

[yhmtsai]

It is OpenMPI 5.0.3 with GCC 13.3.0 and UCX-CUDA
but it only introduces issue with certain #nodes and size.

[pratikvn]

As this is used only when creating the neighborhood communicator, this is probably not a bottleneck and should be fine, but I think the doc needs to be updated. Otherwise, looks good to me.

[pratikvn]

I think the doc of the function needs to be updated ?

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 4 lines in your changes missing coverage. Please review.

Project coverage is 89.65%. Comparing base (d38eea1) to head (1af3592).
Report is 3 commits behind head on develop.

Files with missing lines	Patch %	Lines
core/distributed/neighborhood_communicator.cpp	0.00%	4 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #1842      +/-   ##
===========================================
+ Coverage    89.42%   89.65%   +0.22%     
===========================================
  Files          849      849              
  Lines        71023    70993      -30     
===========================================
+ Hits         63514    63647     +133     
+ Misses        7509     7346     -163     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.