Transfer_ops can now work on both buffers and tensor. Right now, lowering of the tensor case is not supported yet. Differential Revision: https://reviews.llvm.org/D93500
std.tensor_cast
tensor.cast