Enhance PartitionGraph

[ZhennanQin]

Description

Extracted from #14113. This PR covers:

• Add inference_only attr support when SubgraphProperty is created to indicate that this pass should be used for inference only.
• Allow registering multiple subgraph pass within same backend names.
• Refactor the way to run PartitionGraph in simple bind stage to ensure any graph nodes reordering can be correctly handled.

This PR is the full version of #14276.
@xinyu-intel @pengzhao-intel @TaoLv @reminisce @zheng-da

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

• The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
• Changes are complete (i.e. I finished coding on this PR)
• All changes have test coverage:
• Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
• Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
• Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
• Code is well-documented:
• For user-facing API changes, API doc string has been updated.
• For new C++ functions in header files, their functionalities and arguments are documented.
• For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
• Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
• To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

Changes

• Feature1, tests, (and when applicable, API doc)
• Feature2, tests, (and when applicable, API doc)

Comments

• If this change is a backward incompatible change, why must this change be made.
• Interesting edge cases to note here

[anirudhacharya]

@mxnet-label-bot add [pr-awaiting-review]

@ZhennanQin could you please rebase your PR and resolve the conflicts?

[anirudh2290]

Thanks @ZhennanQin . Nice work! Added minor comments, otherwise LGTM.

[pengzhao-intel]

@anirudh2290 thanks a lot for the review :)

@reminisce could you help to review again?

[reminisce]

Some comments on cosmetics. Otherwise LGTM.

[pengzhao-intel]

LGTM. Thanks for your contribution.

Will merge the PR after the CI pass.

[ZhennanQin]

CI is blocked by #14507 and should be fixed by #14504.