distDocker.py fails to ensure a consistent starting state

distDocker.py attempts to reset each node so that it is ready to run autograding images. My understanding is that it does this (in part) by looking for volumes associated with a running autograder; if those exists, it tries to "solve" the problem by killing the docker instance that should be using the volume which should also perform clean up. If an associated container isn't found, however, either those volumes (the directories) should be renamed/removed or at least a warning should occur in the log. This seems like a pushable improvement that would have helped/solved our recent tango issue. Before anyone claims this issue, it might be worth asking if cmu is interested in the fix.