oddity in ordering of coupled transport

[ecoon]

Currently, the expectation of coupled transport is that the subsurface PK is listed first, and then it is hard-coded to always do the surface transport PK first in AdvanceStep()

This is backwards -- if AdvanceStep() on the surface needs to be done first (which is consistent with all other PKs) then it seems that it should be listed first. The potential exception would be if Setup() and/or Initialize() of the subsurface needed to be done first, but this would need to be justified and documented.