refactor: Phase-field and damage solid model cleanup, refactor

[frankfeifan]

To remove dead codes, unify duplicated logic, rename unclear variables, and decouple crack model type from decomposition scheme to enable more options of the phase-field model.

• Remove unnecessary macros in the solver and damage model
• Unify LocalDissipationOption into a constitutive-level definition
• Add FractureModelType (Brittle/Cohesive/Nucleation), and decouple it from decomposition scheme to allow for more phase-field model combinations, e.g.,
  - Brittle + Linear (AT1)/Quadratic (AT2) + No decomposition/Spectral/VolDev
  - Cohesive + Linear + No decomposition/Spectral/VolDev
  - Nucleation + Linear + No decomposition
• Add check for invalid combinations (Nucleation+Spectral, Cohesive+Quadratic, etc.)
• Update existing phase-field xml decks and add new examples
• Enable co-existence of no-phase-field regions (only solid mech, and no damage dof) and phase-field regions. The previous code will complain about empty damage dof set even the regions are not defined in PhaseFieldFracture or PhaseFieldDamageFEM.
• Add viscous regularization for the phase-field equation
• Rebaseline integrated tests

Still wip, and more to come ...

[joshua-white]

Nice cleanup

[frankfeifan]

Added a few well-known benchmark examples in inputFiles/phaseField/benchmark which can be used for future phase-field documentation.

A short report summarizing these examples is attached here:
phase_field_benchmarks_report.pdf

[jhuang2601]

One concern is the size of the meshes for benchmark examples (xx MB), which is too large to be merged into develop repo.
Probably, they should be moved to the GEOSXDATA repo.

Can you also add the coarse version of these meshes for the corresponding smoke tests?

[victorapm]

I see this mesh has about 80k elements. Would it be possible to test the same simulation capabilities on a mesh with 1k elements? This would help speeding up the integrated tests and, at the same time, would solve the issue that Jian pointed out

[frankfeifan]

I see this mesh has about 80k elements. Would it be possible to test the same simulation capabilities on a mesh with 1k elements? This would help speeding up the integrated tests and, at the same time, would solve the issue that Jian pointed out

Thanks Victor. Actually these phase-field simulations have quite stringent mesh requirement. See Fig. 22 in https://onlinelibrary.wiley.com/doi/abs/10.1002/nme.2861

That said, I agree with Jian that move those large mesh files to the GEOSXDATA repo is a good plan. In GEOS, I can create coarse counterparts just for the integrated test purpose.

[victorapm]

That sounds a good plan, thanks, Frank!

[jhuang2601]

see above comment