Fix test-warnings related to symmetry

[maki49]

1. Fix a call of structure symmetrization when symmetry is not analyzed
   In Attention should be paid to some certain PR during integrated testing as they may pose some warning. #3242 @Zhuxuegang2022 mentioned the warning of case 704_LJ_CR introduced in Feature&Refactor: Structure symmetrization during relax #3247, it is because in ESolver_LJ the symmetry does not be analyzed even if symmetry=1 is set in the INPUT file.
   So I changed the condition to call structure symmetrization only when there exist some symmetry operations, i.e. nrotk>0.

2. Change the reference value of some cases: the difference is just because the symmetrization is changed from real space into reciprocal space in Fix: symmetrize charge density in reciprocal space in any case #3081. The old values can be reproduced using real-space symmetrization in my local test.

[dyzheng]

Thanks for your contribution.