[relay] preserve the order of input_info of pytorch

[sweetcocoa]

This is for resolving #14461

The direct cause of this problem is that the order of input nodes found in _analysis.free_vars(ret) is not guaranteed in the order of names in input_info. From what I understand, _analysis.free_vars(ret) uses DFS to find input nodes from output nodes. Since DFS cannot guarantee the order in which nodes are found, it is intuitive to modify the order of input nodes after DFS, as shown in this PR.

[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.

• cc @shingjan _{See #10317 for details}

_{Generated by tvm-bot}

[masahi]

LGTM, please fix lint

[sweetcocoa]

I fixed lint. please re-review.

[masahi]

Some tests are broken, please fix them.

[sweetcocoa]

@masahi I fixed a test failure. However, I cannot interpret the current test results. cross-isa-minimal/pr-head — The build of this commit was aborted. Is this caused by a change in my code?

I checked https://ci.tlcpack.ai/blue/organizations/jenkins/tvm-minimal-cross-isa/detail/PR-14462/6/pipeline/95 but what i see is
not an error message but an interrupt message.

[masahi]

@tvm-bot rerun

[sweetcocoa]

@masahi Could you run again please? (or can i do it myself?)

[masahi]

@tvm-bot rerun

[sweetcocoa]

@masahi Now all checks have passed.