Blksprs

Added

- Add ``repeat`` function

Fixed

- Fix kernels not returning on illegal sparse blocks
- Fix stride not being calculated correctly in some cases

Added

- Add ``build_sparsity_layout_matmul`` function
- Add ``build_sparsity_layout_matmul_fast`` function

Fixed

- Fixed project version number

Added

- Add ``split`` function
- Add ``merge`` function

Fixed

- Fixed ``repeat_interleave`` always using 3 repetitions instead of the specified amount

Added

- Add ``gather_mdi`` and ``scatter_reduce_mdi`` experimental functions

Changed

- Rename ``gather_3d`` to ``gather_mdi``

Added

- Add experimental ``gather_3d`` function

Fixed

- Fixed output dtype differing from input dtype