Suspicious solution in dplasma_dgetrf_1d_mpi test

[G-Ragghianti]

Describe the bug

Post-build test errors on CPU show suspicious solution in dplasma_dgetrf_1d_mpi test:

#14 45.93 Checking the Residual of the solution 
#14 45.93 -- ||A||_oo = 1.025373e+02, ||X||_oo = 4.605746e+00, ||B||_oo= 5.000000e-01, ||A X - B||_oo = 8.500529e+00
#14 45.93 -- ||Ax-B||_oo/((||A||_oo||x||_oo+||B||_oo).N.eps) = 4.284525e+11 
#14 45.93 -- Solution is suspicious !

To Reproduce

Steps to reproduce the behavior:

1. Checkout my docker reproducer at https://github.com/G-Ragghianti/docker
2. cd docker/dplasma
3. docker build --progress=plain -f Docker.redhat9 --build-age BUILD_TYPE=Release .

Expected behavior

No Test errors

Environment (please complete the following information):

• DPLASMA version: ebc116a
• PaRSEC status: external/internal and version: external 4bce29b
• OS: Redhat (rockylinx) 9 and Ubuntu:Noble
• Compiler: gcc11 and gcc13
• MPI version: OpenMPI 5.0.8

Other context

This is reproducible on all systems that I've tested this on. Note that the test docker build also shows a segfault in the dplasma_dpotrf_dtd_untied_mpi test which I will address separately.