Develop training algorithm based on tuning equations in autoregulation model

[gnperdue]

@lucyahuang had the very good idea of making the first on-board model one that implements to the old auto-regulation scheme, but learns the parameters that govern that system.

We can develop an offline version of this model that works with historical data, so we'll start there and think about putting it on a board soon.

[gnperdue]

Lucy has developed the first offline training environment.

[gnperdue]

Oops, this is a duplicate of #29. I'll close that one.

[gnperdue]

The folks at PNNL have basically done this - really what we need at this point is a copy of the code.