Implement a thread-local means to access kernel launch config.#288

tpn wants to merge 4 commits intoNVIDIA:mainNVIDIA/numba-cuda:mainfrom

tpn:280-launch-config-contextvartpn/numba-cuda:280-launch-config-contextvarCopy head branch name to clipboard