It shouldn't really be that hard for them to implement shared rendering of a single scene, it effectively already does it over multiple cores. The queue manager already has the network code to send and receive the required files, it 'just' needs to be able to apportion parts of the image as well.

Alternatively, some python script guru writes us some code that'll automatically do precise area renders of a scene, spit them out to the queue and then a stand alone little app that'll take the renders and stitch them back together again...