Two portions of the KPP interface had bugs
1. the computation of surface buoyancy forcing and friction velocity
were incorrect.  The surfaceBuoyancyForcing subtracted the wrong values
from the heat fluxes from thickness fluxes.  The surfaceFrictionVelocity
had incorrect interpolation to centers

2. the KPP interface for matchBoth is fixed and the indexing is fixed
for matching interior to boundary layer diffusivities.