cuda::launch and launch configuration object with minimal functionali…

…ty (NVIDIA#1950)

* Add the cuda::launch prototype bits

* Some launch prototype cleanup

* Move the test file to new location

* Make launch smoke compile

* Move files around and update header guards

* Remove custom exception and switch to cuda_error

* Minor cleanups

* Add some configuration testing, clean up old test

* Clean up testing and add stream argument to launch

* Add :: before cuda namespace

* Remove explicit exception catching from the tests

* Makefile fix

* Inline docs and cleanups

* Some docs fixes

* Rework config object and add run proof to tests

* Fix includes, level constructor, remove swp file

* Fix level constructor again

* Remove sstream include from a test

* Fix clang format and add comment

* Add missing typename in a cast type

* Fix Windows compilation issues

* More Windows fixes

* Possibly fix Windows warning

* Fix docs typo

Co-authored-by: Bernhard Manfred Gruber <bernhardmgruber@gmail.com>

---------

Co-authored-by: Bernhard Manfred Gruber <bernhardmgruber@gmail.com>