Updates made to the gravity-wave drag parameterization scheme
based on Kim and Arakawa (1995) and Kim (1996).

1) The 'interface height "LINTFC"' - that defines 'low levels' below which the drag can be significantly enhanced if downstream and the flow is sufficiently nonlinear - is now not prescribed but calculated at every grid point (when "LINCLC" = true).

2) The wind tendency at the top of the model "UTGWD(1)" is now extrapolated from the layer below instead of being prescribed. % When the horizontal diffusion is doubled for 4x5 version the GWD scheme is more effective in separating the polar night and subtropical jets.