Should we provide a more robust set of options to the invocation of containers?

Pull request #702, leads me to think that maybe our control over how containers are invoked in CBRAIN is not adequate, and that maybe we should expose such things as "entrypoint" to allow tool developers more control over this?  This may be unnecessary, but I thought I would bring it up. 