Move changes related to CISM's test_coupling into a new testmod dir

Point is to maintain defaultio as just being io-related - not having
extra stuff that changes behavior like this.

testlist_allactive.xml

@@ -195,7 +195,7 @@
       <option name="wallclock"> 00:30 </option>
     </options>
   </test>
-  <test name="NCK_Ld5" grid="f19_g17" compset="B1850GWs" testmods="allactive/defaultio">
+  <test name="NCK_Ld5" grid="f19_g17" compset="B1850GWs" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="cheyenne" compiler="intel" category="prealpha"/>
       <machine name="cheyenne" compiler="intel" category="prebeta"/>
@@ -292,23 +292,23 @@
       <option name="wallclock"> 00:30 </option>
     </options>
   </test>
-  <test name="SMS_Ld5" grid="f19_g17" compset="B1850G" testmods="allactive/defaultio">
+  <test name="SMS_Ld5" grid="f19_g17" compset="B1850G" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="cheyenne" compiler="intel" category="prealpha"/>
     </machines>
     <options>
       <option name="wallclock"> 00:30 </option>
     </options>
   </test>
-  <test name="SMS_Ld5" grid="f09_g17" compset="B1850G" testmods="allactive/defaultio">
+  <test name="SMS_Ld5" grid="f09_g17" compset="B1850G" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="cheyenne" compiler="intel" category="prebeta"/>
     </machines>
     <options>
       <option name="wallclock"> 00:30 </option>
     </options>
   </test>
-  <test name="SMS_Ld5" grid="f19_g17" compset="B1850G1" testmods="allactive/defaultio">
+  <test name="SMS_Ld5" grid="f19_g17" compset="B1850G1" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="cheyenne" compiler="gnu" category="prebeta">
         <options>
@@ -318,7 +318,7 @@
       </machine>
     </machines>
   </test>
-  <test name="SMS_Ld1" grid="f19_g17" compset="B1850G" testmods="allactive/defaultio">
+  <test name="SMS_Ld1" grid="f19_g17" compset="B1850G" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="hobart" compiler="nag" category="prebeta"/>
     </machines>
@@ -334,7 +334,7 @@
   <!--     <option name="wallclock"> 00:30 </option> -->
   <!--   </options> -->
   <!-- </test> -->
-  <test name="SMS_Ld5" grid="f09_g17_gl4" compset="J1850G" testmods="allactive/defaultio">
+  <test name="SMS_Ld5" grid="f09_g17_gl4" compset="J1850G" testmods="allactive/cism/test_coupling">
     <machines>
       <machine name="cheyenne" compiler="gnu" category="prebeta"/>
     </machines>

testmods_dirs/allactive/cism/test_coupling/README

@@ -0,0 +1,2 @@
+The purpose of this testmod directory is to enable CISM's dynamics on
+the short (multi-day) time scales typical of fully-active tests.

testmods_dirs/allactive/cism/test_coupling/include_user_mods

@@ -0,0 +1 @@
+../../defaultio

testmods_dirs/allactive/cism/test_coupling/user_nl_cism

@@ -0,0 +1,4 @@
+
+! This option changes the ice sheet dynamics time step to 1 day rather than 1 year
+! Thus, the ice sheet dynamics can be exercised in a few-day run
+test_coupling = .true.

testmods_dirs/allactive/cism/test_coupling/user_nl_clm

@@ -0,0 +1,12 @@
+
+ ! This is needed to tell CLM to allow the non-annual-boundary glacier changes that arise
+ ! with this testmod.
+ for_testing_allow_non_annual_changes = .true.
+
+ ! When we have daily rather than annual glacier dynamics (as we do in this testmod, due
+ ! to having test_coupling in user_nl_cism), CLM applies the dynbal adjustments in a
+ ! single time step rather than spreading them throughout the year. This can cause
+ ! sensible heat fluxes of thousands of W m-2, which causes CAM's PBL scheme to blow up.
+ ! So force these fluxes to zero for this testmod; this breaks water and energy
+ ! conservation in CLM, but should allow the test to pass.
+ for_testing_zero_dynbal_fluxes = .true.

testmods_dirs/allactive/defaultio/user_nl_cism

@@ -1,6 +1,3 @@
-! This option changes the ice sheet dynamics time step to 1 day rather than 1 year
-! Thus, the ice sheet dynamics can be exercised in a few-day run
-test_coupling = .true.
 
 ! This is needed to give CISM history output in the (typically short)
 ! tests that are done with this testmod directory

testmods_dirs/allactive/defaultio/user_nl_clm

@@ -2,12 +2,3 @@
  hist_ndens     = 1
  hist_nhtfrq    =-24
  hist_mfilt     = 1
- for_testing_allow_non_annual_changes = .true.
-
- ! When we have daily rather than annual glacier dynamics (as we do in this testmod, due
- ! to having test_coupling in user_nl_cism), CLM applies the dynbal adjustments in a
- ! single time step rather than spreading them throughout the year. This can cause
- ! sensible heat fluxes of thousands of W m-2, which causes CAM's PBL scheme to blow up.
- ! So force these fluxes to zero for this testmod; this breaks water and energy
- ! conservation in CLM, but should allow the test to pass.
- for_testing_zero_dynbal_fluxes = .true.