Merge pull request #3996 from RosettaCommons/vmullig/fix_fastrelax_disk_reads
FastRelax repeatedly reads multiple scorefunction weights from disk, making it unusable on large clusters. Fixing this.
Recent changes to FastRelax have it repeatedly reading not only its own relax script, but also scorefunction weights from disk. This means that I can't run anything that uses FastRelax or FastDesign on Mira or Theta, or other giant supercomputers. Fixing this again.
We really need to implement something that firmly prohibits disk reads that are unmanaged by a central manager.