In that case I would catch the error that all deployments are in use and then create one on the fly. I wouldn't even delete it afterward because if the amount of queued jobs has spiked this high, then it could spike this high again
@alien4u has experience with this