Fixing up error handling for fetch_file ops in connection plugins

* enable batch mode (configurable with a config option, on by default)
  for sftp transfers, so we can catch errors more easily
* general cleanup in the local connection plugin and fetch action plugin

Fixes #11612

lib/ansible/plugins/action/fetch.py

@@ -131,9 +131,12 @@ class ActionModule(ActionBase):
             if remote_data is None:
                 self._connection.fetch_file(source, dest)
             else:
-                f = open(dest, 'w')
-                f.write(remote_data)
-                f.close()
+                try:
+                    f = open(dest, 'w')
+                    f.write(remote_data)
+                    f.close()
+                except (IOError, OSError) as e:
+                    raise AnsibleError("Failed to fetch the file: %s" % e)
             new_checksum = secure_hash(dest)
             # For backwards compatibility.  We'll return None on FIPS enabled
             # systems