Optimization rule to merge multiple unions with only constant values

[charygo]

== Test plan ==

The changes are tested with unit tests and by adding a couple of tests with "union all queries over more than one sub-query over constants" to AbstractTestQueries to be run as end-to-end query tests.

Manual tests to generate results and plans are also conducted using the presto cli with tpch database against the presto HiveQueryRunner local service. The outputs are as follows:

presto:sf10> EXPLAIN SELECT * FROM ( select 1, 2, 3 union all select * from (values (5,6,6),(10,20,30)));
Query Plan

• Output[_col0, _col1, _col2] => [expr_19:integer, expr_20:integer, expr_21:integer]
  Estimates: {rows: 3 (45B), cpu: 0.00, memory: 0.00, network: 0.00}
  _col0 := expr_19
  _col1 := expr_20
  _col2 := expr_21
  • Values => [expr_19:integer, expr_20:integer, expr_21:integer]
    Estimates: {rows: 3 (45B), cpu: 0.00, memory: 0.00, network: 0.00}
    (INTEGER 1, INTEGER 2, INTEGER 3)
    (INTEGER 5, INTEGER 6, INTEGER 6)
    (INTEGER 10, INTEGER 20, INTEGER 30)

(1 row)

presto:sf10> SELECT * FROM ( select 1, 2, 3 union all select * from (values (5,6,6),(10,20,30)));
_col0 | _col1 | _col2
-------+-------+-------
1 | 2 | 3
5 | 6 | 6
10 | 20 | 30
(3 rows)

presto:sf10> EXPLAIN SELECT * FROM ((SELECT custkey, nationkey FROM customer ORDER BY custkey LIMIT 10) UNION ALL (SELECT 1, 2) UNION ALL (SELECT 2, 3));
Query Plan

• Output[custkey, nationkey] => [expr_28:bigint, expr_29:bigint]
  custkey := expr_28
  nationkey := expr_29
  • LocalExchange[ROUND_ROBIN] () => [expr_28:bigint, expr_29:bigint]
    • TopN[10 by (custkey ASC_NULLS_LAST)] => [custkey:bigint, nationkey:bigint]
      • LocalExchange[SINGLE] () => [custkey:bigint, nationkey:bigint]
        • RemoteStreamingExchange[GATHER] => [custkey:bigint, nationkey:bigint]
          • TopNPartial[10 by (custkey ASC_NULLS_LAST)] => [custkey:bigint, nationkey:bigint]
            • TableScan[TableHandle {connectorId='tpch', connectorHandle='customer:sf10.0', layout='Optional[customer:sf10.0]'}] => [custkey:bigint, nationkey:bigint]
              Estimates: {rows: ? (?), cpu: ?, memory: 0.00, network: 0.00}
              nationkey := tpch:nationkey
              custkey := tpch:custkey
    • Values => [union_val_expr:bigint, union_val_expr_42:bigint]
      Estimates: {rows: 2 (36B), cpu: 0.00, memory: 0.00, network: 0.00}
      (BIGINT 1, BIGINT 2)
      (BIGINT 2, BIGINT 3)

(1 row)

presto:sf10> SELECT * FROM ((SELECT custkey, nationkey FROM customer ORDER BY custkey LIMIT 10) UNION ALL (SELECT 1, 2) UNION ALL (SELECT 2, 3));
custkey | nationkey
---------+-----------
1 | 2
2 | 3
1 | 15
2 | 13
3 | 1
4 | 4
5 | 3
6 | 20
7 | 18
8 | 17
9 | 8
10 | 5
(12 rows)
`

== RELEASE NOTES ==

General Changes

• Improve/optimize performance for queries that union all over more than one sub-query over constants (values query node).
  This can be enabled/disabled using "optimize_union_over_values" session property.
  The property and unit test portion of the PR have been taken from a previous PR Union All optimisation #15097 (Union All optimisation #15097).

Hive Changes

• NONE

[linux-foundation-easycla]

The committers are authorized under a signed CLA.

• ✅ charygo (9d4f2ba79fb722479e697591c3237c65d8863d32)

[stale]

This pull request has been automatically marked as stale because it has not had recent activity. If you'd still like this PR merged, please comment on the task, make sure you've addressed reviewer comments, and rebase on the latest master. Thank you for your contributions!