Add inelastic strain increment predictor to solid_mechanics materials

[lynnmunday]

Motivation

The previous timesteps inelastic strain increment is a good guess for the current timestep inelastic strain increment and using it would reduce the number of material point Newton solves to converge a nonlinear creep model. Some creep models are also initilialized at the trial stress and it would be good if the trial stress was close was close to the final stress.

Design

Add a stateful material to track the inelastic_strain increment at each time step and use the old value of this as an initialGuess for the next timestep's inelastic strain increment. Also remove this inelastic strain increment prediction from the current step's trial stress to get a better initiliazation of the material model. This change will mostly benefit the Laromance models so changes should be limited to creep model derived classes and not plastic and damage models.

Impact

This should reduce the number of nonlinear material solves for creep models.