remove dependency on Abdul's fork of pyqwalify

[elray1]

We're using a fork of the pyqwalify module that Abdul created; you can see his changes here. This seems awkward (what if that repo disappears?). Can we remove this dependency somehow -- e.g., by extracting the functionality we need, moving the repo over to reichlab, or working with the original pyqwalify code?

I see three main changes in Abdul's code:

1. Look for files that end with .txt instead of just .yml or .yaml. Possible solution: change all the model metadata files in the hub to end with .yml
2. Allow for a silent argument to validation. Possible solution: not sure, but could we somehow sink the log?
3. Return a boolean indicating whether there were any errors rather than the data. There may be no way around this.