[Feature] implement fault permeability models

[jhuang2601]

This PR implements fault permeability models considering permeability change during fault slippage:

• Implement slip dependent permeability model (simple linear multiplier)
• Add documents for SlipDependentPermeability model and CompressibleSolidSlipDependentPermeability
• Add SlipPermeability_pEDFM_smoke to the integratedTests and check the running on Pecan
  GEOSX/integratedTests#178
• Additionally, it modifies the kernels and the solvers such that the dependency of the flux w.r.t to all components of the dispJump is considered (not just the aperture any more) for efem-edfm runs.

Following two models will be implemented in separated PRs

• Willis-Richards model (Barton-Bandis type correlation)
• Input tables for more generalized models

[joshua-white]

I suggest we do this with an L2-norm of the shear displacement instead of the max norm. shearMag = sqrt(u1^2+u2^2). We can then consistently linearize as dShearMag_dDisp = (1/shearMag)*{u1,u2}.

[jhuang2601]

Good idea, please check the update

[joshua-white]

This is a great start, but we should clarify the primary variables and linearization (dOuput_dInput).

[jhuang2601]

This is a great start, but we should clarify the primary variables and linearization (dOuput_dInput).

Thanks for your suggestion. For simplicity, shear displacement dependent permeability is expressed as a function of the dispJump, which denotes the local relative displacement vector in the Embedded Fracture model. And a tanh model is used to linearize the correlation to avoid sharp inflection.

The dispJump has 3 components defined in the local coordinate system:
dispJump[0] : normal component (mechanical aperture)
dispJump[1]: component along tangent direction 1
dispJump[2]: component along tangent direction 2

However, in the Lagrange contact model, a similar vector is used for relative displacement in the local coordinate system but has a different name (localJump). We need to select a universal name for the local relative displacement vector across different fault/fracture models.

[joshua-white]

Nice, changes look great to me. Let's have @CusiniM look quickly to make sure it is all consistent with his framework.

I prefer displacementJump over localJump as nomenclature. It clearly identifies the physical quantity.

[sytuannguyen]

well done!