Modify Stress limiter in VP?

[bobmyhill]

In the visco_plastic rheology, there is an option to approximate plastic yielding with a power law rheology. Currently, the rheology:

• Reuses the reference strain rate used in the first time step as a scaling parameter for the quasiplastic flow law.
• Allows the user to set multiple stress limiter exponents.

The first of these is potentially problematic, because the reference strain rate is first used to initialise the strain rate fields (not affecting the solution), whereas the reuse actually modifies the solution.

The second of these is probably unnecessary complexity - the value of the exponent simply dictates the smoothness of the transition from viscous to plastic.

I'd like to suggest

• adding a parameter "Reference plastic strain rate", and
• replacing "Stress limiter exponents" with "Plastic stress exponent", deprecating the old parameter and enforcing values to be the same. The test suite currently has only one test with multiple Stress limiter exponents, and they are all set to the same value (tests/visco_plastic_complex.prm).

[gassmoeller]

As mentioned in #5946 (comment) I think both are good ideas. It would be great if @anne-glerum or @naliboff could chime in on the usefulness of different stress limiter exponents for different compositions. Is it worth keeping or not?