Update 'Peierls strain rate residual tolerance'

[danieldouglas92]

Following #5338, the default value for the parameter Peierls strain rate residual tolerance seemingly doesn't need to be so strict because the residual is now relative to the log of the strain rate. Opening this PR was motivated by models that I was running using the default value that could not converge, but using a value of 1e-10 - 1e-12 converged and produced negligible differences in the solution. Lowering the value should make models with Peierls creep more stable and faster when using the default value.

This PR updates the documentation of this parameter and changes the default value to mirror the changes in diffusion/dislocation merged in #5338.

[gassmoeller]

This makes sense. Our testers have some problems right now, so this will need to wait a bit until we fix them. One question: #5338 chooses 1e-10 as a threshold, you here choose 1e-12. Any reason for that? Wouldnt it make more sense to use the same threshold?

[danieldouglas92]

I chose 1e-12 because that's what I saw in most of the tests involving Peierls creep, but for consistency I can change it to 1e-10.

[gassmoeller]

/rebuild

[gassmoeller]

The only failing tests are unrelated, so I will merge this. Thanks for the improvement.