Make commuter adrio refactor.

[TJohnsonAZ]

Refactor of ADRIOMakerCensus's _make_commuter_adrio() method to utilize DataFrame.pivot() when converting data into a matrix-like shape.

[JavadocMD]

This is a definite improvement but I feel like we could simplify even further if we can find a way to collapse the county/state cases sooner. Can we return the data from "fetch_commuters()" in a universalized format, such that the adrio no longer cares about the differences between state and county? (Notice that some of the steps are duplicated between the two.)

Also, in "fetch_commuters()" I think we can simplify this logic such that the only cases you need to handle are whether scope.granularity equals state or county, and using scope.get_node_ids() to get away from having to deal with what scope.includes entails. You're accessing too much of the "internals" of these data structures and I think it's making your work harder.

[JavadocMD]

Yes! Perfect.