Bug fix: Make sure planeflight writes out on last timestep of day

Luke Schiferl wrote:

   The planeflight diagnostic in v12.2.0 fails to read out points in
   the final dynamic timestep of the day. It seems that this is due to
   a remnant from before the transition from minutes to seconds used
   to describe timesteps.

This has now been corrected.

Also -- changed the GC version number to 12.3.1 in the file
GeosCore/gc_classic_version.H.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

GeosCore/gc_classic_version.H

@@ -13,4 +13,4 @@
       ! Z = updated for patches that do not break backwards compatibility
       !     with run directories from the last benchmarked version X.Y.0.
       !-----------------------------------------------------------------------
-      GC_CLASSIC_VERSION = '12.3.0'
+      GC_CLASSIC_VERSION = '12.3.1'

GeosCore/planeflight_mod.F

@@ -1711,9 +1711,13 @@ SUBROUTINE PLANEFLIGHT( am_I_Root, Input_Opt,  State_Met,
             PTAUS = PTAUE - ( GET_TS_DIAG() / 3600d0 )
          ! Otherwise write out only half a timestep's worth of data
          ELSE
-            ! Write half hour only if we are on last time step, otherwise
-            ! exit the subroutine
-            IF (MOD(GET_ELAPSED_SEC(), (GET_TS_DIAG()*1440)) == 0) THEN 
+            ! Flush the last timestep to the output file (bmy, 3/28/19)
+            ! NOTE: Luke Schiferl pointed out we should test against
+            ! 86400 seconds instead of GET_TS_DIAG*1440.  GET_TS_DIAG
+            ! by now is already in seconds and 1440 is minutes/day,
+            ! so there's a mismatch.  This caused planeflight data
+            ! for the last timestep of a day not to be written out.
+            IF (MOD(GET_ELAPSED_SEC(), 86400) == 0) THEN
                PTAUS = PTAUE - ( GET_TS_DIAG() / (2.0 * 3600d0) )
             ELSE
                EXIT