Eliminated unnecessary unit conversions for resid_moist etc.

[tbohn]

• [x] closes #849
• [ ] tests passed
• [ ] new tests added
• [ ] science test figures
• [ ] ran uncrustify prior to final commit
• [ ] ReleaseNotes entry

Currently, the soil_con terms max_moist, Wcr, and Wpwp, have units of [mm] but resid_moist has units of [mm/mm]. Every time resid_moist is used in the model physics, it is first multiplied by depth * MM_PER_M for consistency with soil moisture units. Not only is this needless extra computation, but it also makes debugging and modification of the code needlessly difficult. Furthermore, it requires adding depth as an argument to various functions for the sole purpose of appearing in the unit conversion. There are also a few instances in which other soil moisture thresholds have their units converted needlessly (either other variables with the desired units already exist, or the units actually get converted BACK again before use!).

This PR does the following:

• Changes the units of resid_moist to be mm. The conversion from fraction to mm occurs in read_soilparam() and vic_init().
• Replaces all instances in which max_moist is divided by depth * MM_PER_M with porosity.
• Changes the name of soil_con.max_moist_node to soil_con.porosity_node.
• Replaces 160 lines of code in read_soilparam() with a one-line call to existing function soil_moisture_from_water_table().
• These changes are all internal; no changes made to the user interface.

[dgergel]

@tbohn can you post a few figures from when you tested this branch so that we can confirm that results are identical with these changes?

[dgergel]

@tbohn thanks. Once you've posted some plots then this will be good to go.

[tbohn]

Differences in results between this branch and the current develop branch are 0.

1. classic driver:
2. image driver:

[tbohn]

This is the output of gridcell [1,1] in the Stehekin dataset. "Develop" = current develop branch, and "SoilParamUnits" is this PR.

[dgergel]

@tbohn thanks, looks good to me.