Its the latter. The bottleneck here is resource availability. This is a free service, running on Heroku. He probably can't afford to just throw new worker dynos at it.

More about Job queuing: https://devcenter.heroku.com/articles/background-jobs-queuei...