Find the default query plan created for collection and updated it to use
parallelization. The rule searches for unnest followed by an assign for the
collection function expression. When this plan block exists the data source
scan operator added in the blocks place.
Before
plan__parent
UNNEST( $v2 : exp($v1) )
ASSIGN( $v1 : collection( $v0 ) )
ASSIGN( $v0 : constant )
plan__child
After
plan__parent
UNNEST( $v2 : exp($v1) )
DATASCAN( collection( $v0 ) , $v1 )
plan__child
Where DATASCAN operator is configured to use the collection( $v0) for
data represented by the "constant" and $v1 represents the xml document
nodes from the collection.