Add CI for GPU tests on Ruche #4

Job	Run time
lint	22s
check_docker_files	6s
build_base / build_base (openmp, false)	1m 31s
build_base / build_base (cuda, true)	13m 36s
build_base / build_base (hip, false)	6m 28s
build_base / build_base (sycl, false)	15m 51s
build (openmp, gcc, g++, -DKokkos_ENABLE_OPENMP=ON, native)	3m 57s
build (cuda, gcc, g++, -DKokkos_ENABLE_CUDA=ON -DKokkos_ARCH_AMPERE80=ON, native)	7m 30s
build (cuda, gcc, g++, -DKokkos_ENABLE_CUDA=ON -DKokkos_ARCH_AMPERE80=ON, host_device, -DKokkosFF...	7m 17s
build (hip, hipcc, hipcc, -DKokkos_ENABLE_HIP=ON -DKokkos_ARCH_VEGA90A=ON, native)	13m 13s
build (hip, hipcc, hipcc, -DKokkos_ENABLE_HIP=ON -DKokkos_ARCH_VEGA90A=ON, host_device, -DKokkosF...	13m 20s
build (sycl, icx, icpx, -DKokkos_ENABLE_SYCL=ON -DKokkos_ARCH_INTEL_GEN=ON, native)	10m 3s
build (sycl, icx, icpx, -DKokkos_ENABLE_SYCL=ON -DKokkos_ARCH_INTEL_GEN=ON, host_device, -DKokkos...	8m 53s
test (cuda, self-hosted, cuda)	6m 53s
test (openmp, ubuntu-latest)	20s
	1h 49m 20s

Provide feedback