The gene-associations checkouts in Jenkins jobs can cumulatively use too much disk space

[kltm]

The purpose of this issue is to figure out a better way of structuring jobs of the workflow so that redundant multi-gigabyte checkouts are not necessary for a large number of jobs.

Options include keeping some core files in AWS, only checking out the necessary directories, and deleting directories when done.

Will consult with @cmungall and @hdietze when we're all next at the same location.

Until then, will try and clean up problem machines (dicer, maybe grinder) that ran out of disk space.