Merging or joining the streams after splitting

The most effective way to merge streams after splitting is by feeding them into the downstream operator. However, there is a case for which you cannot do this. Consider a slightly modified version of the SplitParallelizer composite that has an output stream (instead of Sink operators) :

namespace sample;
composite SplitParallelizer(output out; input in) {
  param    
    operator $source; 
    operator $body;   
    type     $typeIn;
    type     $typeOut;
  graph

    stream<$typeIn> Src = $source(in) {} 

    (stream<$typeIn> Splitted0   
     <%for(my $i=1; $i<4; ++$i) {%>
       ,stream<$typeIn> Splitted<%=$i%> 
    <%}%>) = ThreadedSplit(Src) {} 
 
    <%for(my $i=0; $i<4; ++$i) {%>   
      stream<$typeOut> Processed<%=$i%>
      = $body(Splitted<%=$i%>)   
    <%}%>

    stream<$typeOut> Processed
     = Filter(Processed0
    <%for(my $i=1; $i<4; ++$i) {%>   
      , Processed<%=$i%>  
    <%}%>) {}   
} 

The preferred operator to merge or join the split streams is a Filter operator. Unlike the Functor and Union operators, the Filter operator does not copy the input tuples; it forwards them to the output stream.