1, it's a logic bug in stun_test
2, there're stun implementation which do not have 'changed address' attribute in response.
In this case, there's not enough evidence to judge NAT type.
@Feng-Zihao Sorry but could you redo this PR against the latest develop branch? The master branch is for releases only. All PRs in the future should go to the develop branch. Thanks!
1, it's a logic bug in stun_test 2, there're stun implementation which do not have 'changed address' attribute in response. In this case, there's not enough evidence to judge NAT type.