Prepull: only one cmssw image is in cache at a time + the prepull takes a longer time than just pulling the CMSSW image at the start of GEN

[mitark]

Currently the prepull step causes disk pressure, so that it deletes other images --> when cmssw8 is pulled, the nodes forget cmssw10 and reco step needs to pull it again. This doesn't happen without the prepull step.

[mitark]

Implementing the prepull steps slowed down the workflow significantly so I commented them out and have not run into problems with image pull limits so remained developing the workflow without the prepull step.