Bug: Additional state variables are cleared at every iteration

[NateAM]

These additional state variables should persist and should be able to be set by the residual classes for output

[NateAM]

Making these a previous iteration storage (even though they aren't) would be a work around. Since they aren't used in the non-linear solve (or rather the user can completely control them) I think this probably makes sense.