It appears that the DataStage Merge stage is just a left outer join with the Master being the "left" side and driving input. The other inputs are joined with the master when possible.
Is that all there is to it? What am I missing?
There are differences - you cannot really compare it to a left outer join. These are some characteristics for the MERGE stage:
it can have a reject link - to be precise it can have one per secondary link as data from the secondary links that did not match get rejected (this is already a huge difference to join)
data from the master link can be dropped (or ignored)
data on the master link need to be duplicate free as data (from the secondray links) is consumed (destructive read)
Please refer to the documentation for more details - but I think it is clear that there are important differences.