Fix underlay.get_logs() method

- use salt minions instead of underlay SSH to get logs from
  all registered nodes
- use IP addresses from underlay.config_ssh for such nodes
  which hostnames not matched any minion
- archive the full dump of logs to /tmp instead of /root
  (archiving to /root caused duplicates in the archive at the
  second time)
- use ${ENV_NAME} as part of the name of the archive in CICD
  jobs

Change-Id: I21e6f5cb8eff6c5bccc707c5c88e509a7bf3a166
diff --git a/jobs/pipelines/swarm-run-pytest.groovy b/jobs/pipelines/swarm-run-pytest.groovy
index 780229d..204aef4 100644
--- a/jobs/pipelines/swarm-run-pytest.groovy
+++ b/jobs/pipelines/swarm-run-pytest.groovy
@@ -74,7 +74,7 @@
                     """)
 
                 def snapshot_name = "test_completed"
-                shared.download_logs("test_completed")
+                shared.download_logs("test_completed_${ENV_NAME}")
                 shared.run_cmd("""\
                     dos.py suspend ${ENV_NAME}
                     dos.py snapshot ${ENV_NAME} ${snapshot_name}
@@ -91,7 +91,7 @@
             common.printMsg("Job is failed", "purple")
             // Downloading logs usually not needed here
             // because tests should use the decorator @pytest.mark.grab_versions
-            // shared.download_logs("test_failed")
+            // shared.download_logs("test_failed_${ENV_NAME}")
             throw e
         } finally {
             // TODO(ddmitriev): analyze the "def currentResult = currentBuild.result ?: 'SUCCESS'"