Use MaterialModelInputs<dim>::reinit() where appropriate.

[ian-r-rose]

This uses the new MaterialModelInputs::reinit() function of @pmbremner in a bunch of places.

[bangerth]

/run-tests

Not all of these places fill all variables, but they will do so now. This may not be what is intended. Specifically, filling the strain rate implies having to compute the viscosity -- which is often expensive. Thoughts?

[ian-r-rose]

Yes, that is a point of contention. I think that the assumption that only viscosity depends on strain rates is not necessarily a good one, and would advocate deprecating it. It would not be hard to add a bool flag to toggle whether to fill the strain rates, which would kick the can on that particular discussion. cc @tjhei @bobmyhill

[tjhei]

Yes, that is a point of contention. I think that the assumption that only viscosity depends on strain rates is not necessarily a good one

We should discuss.

[bobmyhill]

I'm on @ian-r-rose's side.

[bobmyhill]

(but yes, we should discuss!)

[ian-r-rose]

I have restored the ability to select whether or not to compute strain rates, so this should be good to go.

[bangerth]

Outstanding -- thank you! Feel free to open an issue to discuss the question of not always computing the viscosity/not always passing the strain rate.