Fix: optimized printing of start magnetization

[dyzheng]

1. Print starting magnetization of first atom of each element.
2. Print starting magnetization of other atoms which has different starting magnetization with first atom.
3. change OUT(ofs, name, values...) format from "[name] = value1, value2, value3" to "name = [ value1, value2, value3 ]"

Format of starting magnetization would be :

        magnetization of element 2 = [ 0, 0, 0 ]
magnetization of element 2 (atom2) = [ 0, 0, 1 ]

for NSPIN=4 case and

        magnetization of element 1 = 1
magnetization of element 1 (atom2) = -1

for NSPIN=2 case.

[hongriTianqi]

The case tests take 5 minutes more as in PR #2580. We should figure out the reason before accepting this PR.

[hongriTianqi]

The case tests take 5 minutes more as in PR #2580. We should figure out the reason before accepting this PR.

This ISSUE is induced by my own PR #2477. I will take care of this issue.