Fix: let the coverage test pass without checking for its correctness

[WHUweiqingzhou]

Fix #5192

In coverage test, we use

OMP_NUM_THREADS=1 mpirun -np 8 abacus

while in CI tests, we use

OMP_NUM_THREADS=2 mpirun -np 4 abacus

Under these two parallelism, ABACUS gives the result difference above the threshold in 6 cases:

Warning: NG]    1 test cases out of 1547 failed.
1: 102_PW_PINT_UKS
1: 
1: [ERROR     ] 5 test cases out of 1547 produced fatal error.
1: 260_NO_DJ_PK_PU_FM
1: 260_NO_DJ_PK_PU_SO
1: 260_NO_DJ_PK_PU_S1
1: 260_NO_DJ_PK_PU_AFM_URAMPING
1: 304_NO_GO_AF_atommag

In fact, CI testing ensures the correctness of the code, while coverage detection may not check for correctness.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 74.18%. Comparing base (146a509) to head (f9b0101). Report is 69 commits behind head on develop.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## develop #5469 +/- ## =========================================== + Coverage 74.08% 74.18% +0.09% =========================================== Files 1118 1123 +5 Lines 105254 105727 +473 Branches 14578 14627 +49 =========================================== + Hits 77978 78432 +454 + Misses 21355 21249 -106 - Partials 5921 6046 +125 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.