Transposed 2D Load

Hi, the async_work_group_copy_2D2D loads rectangular tiles from global to local mem or vice versa, is there a function (potentially from another extension) that acts like async_work_group_strided_copy_2D2D to transpose the src rectangle?

Start by comparing the async_work_group_copy_2D2D and async_work_group_strided_copy_2D2D entry points mentioned in the issue, then check the OpenCL extension specifications for a transposing variant. Done means establishing whether such an operation exists and documenting the relevant function or the missing capability.

Written by the indexing model from the issue text.