Don't use super() when normalizing username using PAM

- We don't need the extra normalization of that function.
- Also add in username_map support here.  It probably isn't needed
  most of the time with PAM, but it keeps things consistent and is
  easier than documenting an exception.

jupyterhub/auth.py

@@ -787,6 +787,8 @@ class PAMAuthenticator(LocalAuthenticator):
             import pwd
             uid = pwd.getpwnam(username).pw_uid
             username = pwd.getpwuid(uid).pw_name
+            username = self.username_map.get(username, username)
+        else:
             return super().normalize_username(username)
 
 class DummyAuthenticator(Authenticator):