Gemm impl compilation errors (OpenMP backend) in some nightlies - Power8+gcc7.2 w/cpp17, Clang10

[ndellingwood]

Following merge of #935 the following nightly builds began failing:

1. White testbed - OpenMP_gcc_720_cpp17

Error messages:

23:36:25 In file included from /home/jenkins/white/workspace/KokkosKernels_White_OpenMP_gcc_720_cpp17/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_spec.hpp:52:0,
23:36:25                  from /home/jenkins/white/workspace/KokkosKernels_White_OpenMP_gcc_720_cpp17/TestAll_2021-06-01_23.31.54/gcc/7.2.0/OpenMP-release/src/impl/generated_specializations_cpp/gemm/Blas3_gemm_eti_DOUBLE_LAYOUTLEFT_EXECSPACE_OPENMP_MEMSPACE_HOSTSPACE.cpp:48:
23:36:25 /home/jenkins/white/workspace/KokkosKernels_White_OpenMP_gcc_720_cpp17/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_impl.hpp: In lambda function:
23:36:25 /home/jenkins/white/workspace/KokkosKernels_White_OpenMP_gcc_720_cpp17/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_impl.hpp:438:3: sorry, unimplemented: unexpected AST of kind omp_simd
23:36:25    });
23:36:25    ^
23:36:25 /home/jenkins/white/workspace/KokkosKernels_White_OpenMP_gcc_720_cpp17/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_impl.hpp:438: confused by earlier errors, bailing out
23:36:25 make[2]: *** [src/CMakeFiles/kokkoskernels.dir/impl/generated_specializations_cpp/gemm/Blas3_gemm_eti_DOUBLE_LAYOUTLEFT_EXECSPACE_OPENMP_MEMSPACE_HOSTSPACE.cpp.o] Error 1

Reproducer instructions:

module load gcc/7.2.0 cmake/3.19.3

$KOKKOSKERNELS_PATH/cm_generate_makefile.bash --with-devices=OpenMP --arch=Power8 --compiler=g++ --cxxflags="-O3 -Wall -Wunused-parameter -Wshadow -pedantic -Werror -Wsign-compare -Wtype-limits -Wignored-qualifiers -Wempty-body -Wclobbered -Wuninitialized " --cxxstandard="17" --kokkos-path=$KOKKOS_PATH --kokkoskernels-path=$KOKKOSKERNELS_PATH --with-scalars='double,complex_double' --with-ordinals=int --with-offsets=int,size_t --with-layouts=LayoutLeft --no-examples 

2. kokkos-dev-2 testbed - clang/10 with OpenMP

Error messages:

06:05:05 In file included from /home/jenkins/slave/workspace/KokkosKernels_KokkosDev2_CLANG10/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_spec.hpp:52:
06:05:05 /home/jenkins/slave/workspace/KokkosKernels_KokkosDev2_CLANG10/kokkos-kernels/src/blas/impl/KokkosBlas3_gemm_impl.hpp:411:63: error: loop not vectorized: the optimizer was unable to perform the requested transformation; the transformation might be disabled or specified as part of an unsupported transformation ordering [-Werror,-Wpass-failed=transform-warning]
06:05:05   Kokkos::parallel_for(Kokkos::TeamThreadRange(team,blockA0), [&] (const int i) {

Reproducer

module load sems-clang/10.0.0 sems-cmake/3.17.1
module swap sems-gcc/5.3.0 sems-gcc/8.3.0

$KOKKOSKERNELS_PATH/cm_generate_makefile.bash --with-devices=OpenMP --arch=SNB,Volta70 --compiler=clang++ --cxxflags="-O3 -Wall -Wunused-parameter -Wshadow -pedantic -Werror -Wsign-compare -Wtype-limits -Wuninitialized " --cxxstandard="14" --kokkos-path=$KOKKOS_PATH --kokkoskernels-path=$KOKKOSKERNELS_PATH --with-scalars='double,complex_double' --with-ordinals=int --with-offsets=int,size_t --with-layouts=LayoutLeft 

@e10harvey do you have time to take a look?