un-aliased subquery is supported by Spark SQL for a long time. Its semantic was not well defined and has confusing behaviors, and it's not a standard SQL syntax, so we disallowed it in https://issues.apache.org/jira/browse/SPARK-20690 .
However, this is a breaking change, and we do have existing queries using un-aliased subquery. We should add the support back and fix its semantic.
After the fix, there is no syntax change from branch 2.2 to master, but we invalid a weird use case:
SELECT v.i from (SELECT i FROM v). Now this query will throw analysis exception because users should not be able to use the qualifier inside a subquery.