mpi: do node-exclusive scheduling for cray-pals PMI

Problem: as documented in the "CORAL2: Flux on Cray Shasta" page
in the flux docs, two flux subinstances sharing the same nodes
can fail due to overlapping port numbers. For some reason, ever
since the vcpu test was added, this has been happening more often.

The solution is to do node-exclusive scheduling at the top level
so the jobs run sequentially.

mpi/outer_script.sh

@@ -35,7 +35,7 @@ COMPILERS="${LCSCHEDCLUSTER}_COMPILERS"
 for mpi in ${!MPIS}; do
     for compiler in ${!COMPILERS}; do
         if [[ $mpi == "cray-mpich" ]]; then
-            EXTRA_FLUX_SUBMIT_OPTIONS="-o pmi=cray-pals" flux batch -N2 -n4 --flags=waitable --output=kvs $MPI_TESTS_DIRECTORY/inner_script.sh $compiler $mpi
+            EXTRA_FLUX_SUBMIT_OPTIONS="-o pmi=cray-pals" flux batch --exclusive -N2 --flags=waitable --output=kvs $MPI_TESTS_DIRECTORY/inner_script.sh $compiler $mpi
         elif [[ $mpi == "openmpi"* ]]; then
             EXTRA_FLUX_SUBMIT_OPTIONS="-o pmi=pmix" flux batch -N2 -n4 --flags=waitable --output=kvs $MPI_TESTS_DIRECTORY/inner_script.sh $compiler $mpi
         else