Rework TDyDriver* into its own data structure and interface

[jeff-cohere]

In recent days, we've been discussing how the TDycore library interface should look to a "client". I think we're converging on a system where you create and configure a dycore object, extract a DM from it, and pass that DM to a solver of your choice to use for your simulation (see PR #211).

We also discussed a higher-level "black box" interface that doesn't require a user to manipulate PETSc's solvers. I think te TDyDriver functions represent the beginning of this higher-level interface. However, I'd like to devise a separate data structure called TDyDriver that has its own functions, rather than have the TDy structure handle both the high- and low-level interfaces.

@bishtgautam @ghammond86