Cleanup: make ET partitioning more explicit

[tbohn]

• Version: VIC 5.0.1 develop branch
• Problem: the logic in computing soil evaporation (esoil) is not very clean nor does it make clear which portion of ET comes from esoil, until the partitioning is cleared up in put_data().
• Another problem: the default value of RARC_SOIL (architectural resistance for soil evaporation) is currently set to 100 s/m. Evaluation of VIC performance at Ameriflux towers suggests a more optimal value would be 250 s/m.
• Solution: (a) introduce explicit esoil, transp, and evap terms into the layer struct; (b) remove the bare_evap_frac term; (c) modify logic in func_surf_energy_bal() and other functions to use these terms; (d) make the logic cleaner; and (e) change the default value of RARC_SOIL to 250.0 s/m.

[tbohn]

I've broken out the change to the default SOIL_RARC value into another issue (#767). I subsequently closed that issue. It turns out that there are other issues impacting soil evaporation. Trying to optimize SOIL_RARC without addressing these other issues might give us the right answer for the wrong reason (compensating errors). These other issues need to be addressed first, before attempting to optimize SOIL_RARC.

[tbohn]

Addressed in #777