Unfortunately, this required a taskprocessor listener change that makes listener allocation
utterly silly. I'm going to change the scheme so that allocation of taskprocessor listeners
is done internally within taskprocessor code. This will make it parallel with threadpool
code, which is a good thing.