Reduce the size of barotropic diagnostic arrays

  Restrict the size of 8 barotropic diagnostics to not have wide halos as this
led to unnecessary copies, and avoid the use of duplicated wide-halo arrays for
the calculation of uhbtav, CS%ubtav, vhbtav and CS%vbtav, thereby eliminating 4
unneeded arrays.  The code setting ubt_int, uhbt_int, vbt_int and vhbt_int was
moved into apply_u_velocity_OBCs and apply_v_velocity_OBCs, eliminating a pair
of loops that look for where OBCs are applied, so that there are now just two
such loops for each of the u- and v-velocities inside of the barotropic time
stepping loop (ideally there would just be one).  The now unused routines
store_u_for_OBCs and store_v_for_OBCs were eliminated.   All answers are bitwise
identical, and no interfaces are changed.

Author: Hallberg-NOAA