LALSimulation: test_IMRPhenomXPHM fails on two different architectures.

test_IMRPhenomXPHM fails at the same assert:

https://git.ligo.org/lscsoft/lalsuite/-/blame/master/lalsimulation/test/python/test_phenomX.py#L186

In two different scenarios:

1. @alexander.pace discovered it when building MKL-optimized rpms in an el7 container in gitlab CI (affected pipeline: https://git.ligo.org/alexander.pace/gstlal-dev/-/jobs/861843)

2. @steffen.grunewald encountered it when building arm64 debian packages.