MueLu: anisotropic scaling test stagnates for OpenMP

[jhux2]

Bug Report

One of MueLu's scaling tests stagnates for OpenMP vs Serial node. For OpenMP, the test converges in 190 iterations, whereas for Serial node the test converges in ~66 iterations.

To run the test:
mpirun -np 4 MueLu_Driver.exe --xml=anisotropic.xml --tol=1e-6 --its=66

PR results: https://trilinos-cdash.sandia.gov/test/41212727

Related to #12533.

[github-actions]

Automatic mention of the @trilinos/muelu team

[jhux2]

Probably related to #12351. @GrahamBenHarper

[GrahamBenHarper]

Also worth noting the Cuda build only does a hair better https://trilinos-cdash.sandia.gov/test/41214385
Based on the results, I'm assuming the majority of this difference is caused by BuildAggregatesRandom?

[github-actions]

This issue has had no activity for 365 days and is marked for closure. It will be closed after an additional 30 days of inactivity.
If you would like to keep this issue open please add a comment and/or remove the MARKED_FOR_CLOSURE label.
If this issue should be kept open even with no activity beyond the time limits you can add the label DO_NOT_AUTOCLOSE.
If it is ok for this issue to be closed, feel free to go ahead and close it. Please do not add any comments or change any labels or otherwise touch this issue unless your intention is to reset the inactivity counter for an additional year.