add more i16/u16 tests; fix cuda align for i16/u16

[kawogi]

This is a follow-up for #439, adding the missing test cases (and some more).

Also the CUDA-alignment has been fixed (to my knowledge).

Test-coverage should at least be back to normal. API is backwards-compatible (no version-bump required IMO).

[bitshifter]

This is the alignment table I've been using, https://docs.nvidia.com/cuda/archive/11.7.0/cuda-c-programming-guide/index.html#vector-types__alignment-requirements-in-device-code. I think what you have looks about right.

Thanks for all your work here!

[kawogi]

Thanks for checking. I'm happy to give something back and your crate was super easy to work with.