Use CMake and upgrade to cudawrappers 0.6.0

This is a follow-up on !3 (closed), rebased onto the latest master and refactored to work with cudawrappers 0.6.0 (which is now header-only). An initial CI setup has also been added.

cmake/tensor-core-correlator-config.cmake.in

+include(CMakeFindDependencyMacro)
+
+find_package(CUDAToolkit @CUDA_MIN_VERSION@ REQUIRED)
+
+foreach(component ${@PROJECT_NAME@_FIND_COMPONENTS})
+  include(${CMAKE_CURRENT_LIST_DIR}/${component}-config.cmake)
+endforeach()