Fix bad xgrid along zero longitude when LND grid not same as ATM (#379)

- Address issue #366
- When LND grid is not the same as ATM the xgrid is wrong along the Prime Meridian,
  E.g., if ATM is c96 and LND is c48, the xgrid cell areas are small by almost a factor of 2 along longitude=0
  and the land mask has a spurious line along longitude=0
- The root cause is the Ocean grid is shifted right by 360 degree assuming that the land and atmos grids are the same
  which cause the atmXlnd exchange grid cells that are just to the left of the longitude=0 to be ignored.
- The fix is to (re)apply the fixlon to shift Ocean grid based on the atmxlnd_x[l][0] rather than the atm_avg.

Author: nikizadehgfdl

src/make-coupler-mosaic/make_coupler_mosaic.c

@@ -1662,7 +1662,15 @@ int main (int argc, char *argv[])
 	    if(lnd_frac > MIN_AREA_FRAC) { /* over land */
 	      /* find the overlap of atmxlnd and ocean cell */
 	      for(l=0; l<count; l++) {
-		if( clip_method == GREAT_CIRCLE_CLIP )
+            /*Apply a longitude fix to ocean cell xo based on the center of the atmxlnd cell
+            Otherwise, when lnd and atm grids are not the same some exchange grids may be missed, particularly around the Prime Meridian lon=0 
+            */
+            if(!lnd_same_as_atm){
+              no_in = fix_lon (xo, yo, 4, atmxlnd_x[l][0]); //Shifts xo so the center to be within atmxlnd_x[l][0]-pi to atmxlnd_x[l][0]+pi, 
+              xo_min = minval_double (no_in, xo);
+              xo_max = maxval_double (no_in, xo);		
+           }
+    if( clip_method == GREAT_CIRCLE_CLIP )
 		  n_out = clip_2dx2d_great_circle(atmxlnd_x[l], atmxlnd_y[l], atmxlnd_z[l], num_v[l], xo, yo, zo, 4,
 						  x_out, y_out, z_out);
 		else {