Issue: Basis selection for ODE Integrator in Splitting Cases

[bornhoft]

When we solve the dU_q/dt = S(U_q) system of ODEs for splitting cases we are assuming that an orthogonal basis has been utilized allowing us to directly solve this equation in an uncoupled manner. In practice, this seems to work ok for non-orthogonal basis (like Lagrange) but there is likely errors being introduced because we haven't first projected onto an orthogonal basis first. This needs to be addressed at some point or a flag needs to be added to not allow non-orthogonal basis with splitting operations.