[FEA]: Improve device hierarchy queries

[davebayer]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this request and that I agree to the Code of Conduct

Area

libcu++

Is your feature request related to a problem? Please describe.

Current way of using hierarchy queries is:

cuda::hierarchy_dimensions<...> dims;
dims.extents(cuda::block, cuda::grid);
dims.index(cuda::block, cuda::grid);

which is a bit hard to read and can look weird for the .index(...) for example.

Describe the solution you'd like

I'd like to do something that is more natural to a CUDA user:

cuda::hierarchy_dimensions<...> dims;
cuda::block.extents(cuda::grid, dims);
cuda::thread.index(cuda::grid, dims);

I think it's easier to read: block index in grid for given dims

The nice thing about this approach is that a user needn't to use hierarchy dimensions at all and do:

cuda::block.extents(cuda::grid);
cuda::thread.index(cuda::grid);

I will also provide .meow_as<int>(...) variants which can override the type that is returned by the query.

Describe alternatives you've considered

No response

Additional context

No response