NiO 512 atom scaling run failure on Frontier

**Describe the bug**
Frontier/Crusher issue OLCFHELP-11849
Running NiO 512 atom scaling runs.
1. branching is disabled.
2. matrix inversion on host. because https://github.com/QMCPACK/qmcpack/issues/4555

Not all runs are successful. Failures are related to #4553.
I ran 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048, 4096 nodes.
Got failure on 1024 and 4096 node runs. All the rest runs have good linear scaling (flat timing).

**To Reproduce**
reproducer input/output files.
[dmc-a512-scaling-failed-runs.zip](https://github.com/QMCPACK/qmcpack/files/11226683/dmc-a512-scaling-failed-runs.zip)

Download hdf5 from https://github.com/QMCPACK/qmcpack/blob/9fe29deb747c86b8f9587ffe8e4a5eeee7fa27fe/tests/performance/NiO/README#L59
pseudopotential files can be found at
https://github.com/QMCPACK/qmcpack/tree/develop/tests/pseudopotentials_for_tests
Ni.opt.xml and O.ncpp.xml (need to be renamed to O.xml).

Froniter/Crusher build recipe, using rocm 5.4.3
https://github.com/QMCPACK/qmcpack/blob/develop/config/build_olcf_frontier_ROCm.sh
I was testing commit dbe06417bce423febef09c3fd2a36409c79be866 but 5d7a24f23b87151aaf56fab7e247cbf0e9398080 a later commit with the build recipe should also reproduce the issue.

Start by unpacking dmc-a512-scaling-failed-runs.zip and reviewing the NiO scaling README, the Frontier ROCm build recipe, and the tested commits. Reproduce the 4–4096 node scaling runs, focusing on the failures at 1024 and 4096 nodes; done means identifying why those runs fail and documenting or correcting the failure so the scaling run completes.

Written by the indexing model from the issue text.