Before you delete again, can you create a gist of job yaml so we can see what's exploded?
- lastProbeTime: 2016-01-08T18:48:51Z
Try `oc delete job/thejob --cascade=false --loglevel=6`. That should avoid any reaping waits and provide some debug info if it still doesn't work. Once that happens, you will be left with cleaning up the rest of what the job created by hand.