Add CI for GPU tests on Ruche #2

Job	Run time
check_docker_files	3s
lint	10s
build_base / build_base (openmp, false)	1m 58s
build_base / build_base (cuda, true)	8m 23s
build_base / build_base (hip, false)	3m 52s
build_base / build_base (sycl, false)	3m 1s
build (openmp, gcc, g++, -DKokkos_ENABLE_OPENMP=ON, native)	3m 23s
build (cuda, gcc, g++, -DKokkos_ENABLE_CUDA=ON -DKokkos_ARCH_AMPERE80=ON, native)	7m 14s
build (cuda, gcc, g++, -DKokkos_ENABLE_CUDA=ON -DKokkos_ARCH_AMPERE80=ON, host_device, -DKokkosFF...	6m 56s
build (hip, hipcc, hipcc, -DKokkos_ENABLE_HIP=ON -DKokkos_ARCH_VEGA90A=ON, native)	12m 10s
build (hip, hipcc, hipcc, -DKokkos_ENABLE_HIP=ON -DKokkos_ARCH_VEGA90A=ON, host_device, -DKokkosF...	12m 2s
build (sycl, icx, icpx, -DKokkos_ENABLE_SYCL=ON -DKokkos_ARCH_INTEL_GEN=ON, native)	10m 13s
build (sycl, icx, icpx, -DKokkos_ENABLE_SYCL=ON -DKokkos_ARCH_INTEL_GEN=ON, host_device, -DKokkos...	10m 27s
test (cuda, self-hosted, cuda)	25m 50s
test (openmp, ubuntu-latest)	19s
	1h 46m 1s

Provide feedback