Updates to SGH model to enhance channel stability

[alexolinhager]

Introduces updates to the subglacial hydrology model to make it more stable, particularly around irregular bed/surface topography, near domain edges, over subglacial lakes, at the grounding line when coupled with ice dynamics, and when using the channelPressureFreeze term.

Important changes made:

• Add missing channel melt source term to mass conservation equation
• Disable diffusive flux at grounding line and at boundary edges
• Update of gradients along boundaries of domain
• Update gradMagPhiEdge to be dependent on full hydropotential, used only by channels. Distributed system is now dependent on gradMagPhiBaseEdge, which is equivalent to gradMagPhiEdge in previous commits.
• Introduces iceThicknessHydro and upperSurfaceHydro. These are copies of thickness and upperSurface that have local minima/maxima made equal to the mean of their neighbors. Done to enhance channel stability and is only used by the hydro model.
• New output variable hydroTerrestrialMarginMask
• Allow for waterPressure above overburden (but disallow negative effective pressure)
• Impose small outflowing waterflux and hydropotential gradient at grounding line. Used instead of zeroing out waterFlux to inhibit inflow.
• Give ocean zero hydropotential and waterPressure, then allow for two-sided calculation of hydropotential across grounding line
• Introduce waterPressureSmooth, a smoothed version of waterPressure that is used only when calculating waterPressureSlopeNormal and channelPressureFreeze. Helps keep channelPressureFreeze stable
• Create option to limit channel growth over subglacial lakes. Done by making channelDischarge = 0 when subglacial lake depth is over a threshold value or if waterPressure is over flotation. channelArea is not affected so that channels do not have to start from scratch if lake geometry changes, essentially creating "ghost channels" over lakes

[alexolinhager]

To be replaced by #106 and #118 but keeping open for reference until those are complete.

[matthewhoffman]

@alexolinhager , can we close this now that #106 and #118 are merged?

[alexolinhager]

Yes this should be ready to close. I'll do that now