This worklog extends the support for transformations of subqueries
to derived tables with joins for correlated scalar subqueries.

For an eligible correlated scalar subquery, the subquery is
converted into a derived table with grouping added on
the non-correlated field of the predicate. The correlated
predicate is lifted out from the derived table and the
non-correlated field is replaced with the field in the
derived table. The derived table is outer joined on the
lifted predicate in order to select the right group to
match up the outer row. A cardinality check is also added
to see if any of the groups in the derived table produced
more than one row.

The transformation is controlled by the the optimizer switch
"subquery_to_derived". It is on by default for secondary
engines like RAPID. But could be enabled for primary engines
like Innodb.

For further details, see the worklog page.

Author: Dag Wanvik <dag.wanvik@oracle.com>
Approved by : Roy lyseng <roy.lyseng@oracle.com>