Provide more extensive support for collapse into padded tensor in codegen or generalize checks for when we can safely add pad encodings

[jtuyls]

Sometimes when we add pad encodings, the blocked dimensions can make it hard to fold the tensor.collapse_shape and the iree_tensor_ext.dispatch.tensor.store operations, for example:

%collapsed = tensor.collapse_shape %36 [[0], [1, 2, 3]] : tensor<4x8x4x128xf8E4M3FNUZ> into tensor<4x4096xf8E4M3FNUZ>
iree_tensor_ext.dispatch.tensor.store %collapsed, %21, offsets = [0, 0], sizes = [4, 4096], strides = [1, 1] : tensor<4x4096xf8E4M3FNUZ> -> !iree_tensor_ext.dispatch.tensor<writeonly:tensor<4x4224xf8E4M3FNUZ>>

Currently, codegen only handles the case when the padded dimension is a multiple of the blocked dimensions. This needs some investigation on how to best generalize this towards more cases and/or add checks to avoid getting into this case.