Revert "[sparse] semi-structured sparse + torch.compile support (#111049)"
This reverts commit 408f210938176870133a3dde5e8fbc4926cafbc0.
Reverted https://github.com/pytorch/pytorch/pull/111049 on behalf of https://github.com/clee2000 due to Sorry I'm pretty sure this caused a memory leak https://hud.pytorch.org/pytorch/pytorch/commit/408f210938176870133a3dde5e8fbc4926cafbc0 https://github.com/pytorch/pytorch/actions/runs/6550388354/job/17790615103 `test_sparse_semi_structured.py::TestSparseSemiStructuredCUDA::test_mlp_contiguous_relu_compile_backend_cutlass_dense_input_shape_(1, 128)_cuda - RuntimeError: CUDA driver API confirmed a leak in __main__.TestSparseSemiStructuredCUDA.test_mlp_contiguous_relu_compile_backend_cutlass_dense_input_shape_(1, 128)_cuda! Caching allocator allocated memory was 235008 and is now reported as 352256 on device 0. CUDA driver allocated memory was 359333888 and is now 361431040.` ([comment](https://github.com/pytorch/pytorch/pull/111049#issuecomment-1767186569))