increase relative tolerance for 32 test of test_IMRPhenomXPHM

The MR raises the tolerance for the 32-mode test in test_IMRPhenomXPHM_SpinTaylor, as implemented in lalsimulation/test/python/test_phenomX.py. The current tolerance appears too tight for arm64-Debian Buster, as discussed in this issue. No other changes have been made.

fixes: #669 (closed)

lalsimulation/test/python/test_phenomX.py

@@ -406,7 +406,7 @@ def test_IMRPhenomXPHM_SpinTaylor():
     new_result  =  np.array(gen_test_data(0.5, lalsimulation.IMRPhenomXPHM, [[3,2]], lalparams=lalDict, generic_spins=True,release=122019))
 
     # rtol with 32 mode needs to be more lenient
-    np.testing.assert_allclose(new_result, expected_result, rtol=3.1e-4, err_msg="IMRPhenomXPHM 32 mode test failed")
+    np.testing.assert_allclose(new_result, expected_result, rtol=5e-4, err_msg="IMRPhenomXPHM 32 mode test failed")
 
 
 def test_IMRPhenomXAS_NRTidalv2():