UV_LOGDRAG ignored when SOLVE3D undefined

[m.hadfield]

Just now I ran a 2D barotropic simulation with logarithmic bottom stress and was surprised to find the bottom stresses were all zero. I shouldn't have been surprised, because it's clear from set_vbc.F that there is no code for UV_LOGDRAG when SOLVE3D is undefined.

Surely a logarithmic bottom-stress formulation is still meaningful for a 2D simulation: the height in the logarithmic formula would just be 0.5*(h+zeta). (I realise this is not exact, but it would have the desired effect of reducing the effective bottom drag coefficient as the water depth increases.) So I propose to add the code to support this combination. What do people think?

Alternatively, I propose to modify the code in checkdefs.F, so that it stops if UV_LOGDRAG is defined and SOLVE3D isn't.