[Quantization]: Update simulated_quantize to infer correct layout

[f2013519]

In our BYOC flow, we invoke the Convert Layout pass after simulated_quantize annotations are inserted in the IR.

However, after Convert Layout, we observe layout transforms to the original layout are inserted before the simulated_quantize annotations, as the compiler is currently unable to infer the layouts from the input.

This PR allows propagation of the input layouts to avoid the insertion of additional layout transforms.

[tvm-bot]

Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.

• No users to tag found in teams: quantization _{See #10317 for details}

_{Generated by tvm-bot}

[f2013519]

cc: @junrushao @masahi @vinx13