Update GFS_debug (GFS_interstitialtoscreen)

[climbfuji]

Description of Changes:

Update GFS_debug schemes: remove GFS_interstitialtoscreen_timestep_init and GFS_interstitialtoscreen_init. These schemes never made sense, because the contents of the GFS interstitial DDTs aren't valid for these phases (only valid for the run phase).

Further, with the upcoming changes by @dustinswales that change when and how GFS_interstitial is allocated (triggered by ccpp-framework changes), they would create segmentation faults.

Tests Conducted:

There is currently on regression test in the ufs-weather-model that exercises the GFS_debug schemes.

I modified the control_p8_intel test to run the GFS_debug schemes as follows:

[Dom.Heinzeller@ufe04 UFSATM]$ git diff
diff --git a/ccpp/suites/suite_FV3_GFS_v17_p8_ugwpv1.xml b/ccpp/suites/suite_FV3_GFS_v17_p8_ugwpv1.xml
index 6d32a42a..d55a4004 100644
--- a/ccpp/suites/suite_FV3_GFS_v17_p8_ugwpv1.xml
+++ b/ccpp/suites/suite_FV3_GFS_v17_p8_ugwpv1.xml
@@ -4,6 +4,7 @@
   <!-- <init></init> -->
   <group name="time_vary">
     <subcycle loop="1">
+      <scheme>GFS_diagtoscreen</scheme>
       <scheme>GFS_time_vary_pre</scheme>
       <scheme>GFS_rrtmg_setup</scheme>
       <scheme>GFS_rad_time_vary</scheme>
@@ -12,6 +13,8 @@
   </group>
   <group name="radiation">
     <subcycle loop="1">
+      <scheme>GFS_diagtoscreen</scheme>
+      <scheme>GFS_interstitialtoscreen</scheme>
       <scheme>GFS_rrtmg_pre</scheme>
       <scheme>GFS_radiation_surface</scheme>
       <scheme>rad_sw_pre</scheme>

I ran this test with -DDEBUG=ON and saved the run directory on Ursa in a different path to show that the run completed successfully (out) and that the GFS_debug output is written to err as expected: /scratch4/NCEPDEV/stmp/Dom.Heinzeller/RT_RUNDIRS_KEEP_ufspr2880

The failing GitHub actions tests have nothing to do with my changes.

This PR was also tested with NOAA-EMC/ufsatm#1004 and ufs-community/ufs-weather-model#2880.

Dependencies:

n/a

Documentation:

n/a

Issue (optional):

n/a

Contributors (optional):

n/a

[dustinswales]

@climbfuji Thanks for making these changes.
Looks good to me.