fix(query): preserve nullable reflexive equality filters

[youngsofun]

I hereby agree to the terms of the CLA available at: https://docs.databend.com/dev/policies/cla/

Summary

• Fixes bug: HAVING (AGG(col) = AGG(col)) keeps groups whose aggregate is NULL — the same self-comparison in a projection column filters them correctly #20379.
• Preserve nullable reflexive equality predicates during filter inference so NULL semantics are not lost.
• Continue eliminating non-nullable reflexive equalities and keep the existing inference path unchanged for other equality predicates.
• Add a named optimizer replay case with raw, optimized, and physical goldens to explicitly protect elimination of non-nullable x = x.
• Update the TPC-DS Q01 logical and physical optimizer snapshots to retain the nullable self-equality filter pushed to the scan.

Tests

• Unit Test
• Logic Test
• Benchmark Test
• No Test - Explain why

Validation performed:

• cargo fmt --all -- --check
• TEST_SUBDIR=regressions cargo test --package databend-common-sql --test it -- planner::test_lite_replay_service_optimizer_cases --exact --nocapture
• TEST_SUBDIR=regressions cargo test --package databend-query --test it -- sql::planner::optimizer::optimizer_test::test_optimizer --exact --nocapture
• TEST_SUBDIR=tpcds cargo test --package databend-common-sql --test it -- planner::test_lite_replay_service_optimizer_cases --exact --nocapture
• cargo clippy --package databend-common-sql --tests -- -D warnings

Type of change

• Bug Fix (non-breaking change which fixes an issue)
• New Feature (non-breaking change which adds functionality)
• Breaking Change (fix or feature that could cause existing functionality not to work as expected)
• Documentation Update
• Refactoring
• Performance Improvement
• Other (please describe):

Risk and performance

The optimizer only retains the original x = x predicate when x is nullable. Non-nullable self-equalities still take the existing elimination path, and other equality inference is unchanged. Retaining the nullable predicate can also push the required NULL rejection closer to a scan.

AI assistance

• AI usage: A coding agent diagnosed the filter-inference bug, drafted the minimal optimizer fix, investigated affected optimizer snapshots, and added logic and optimizer regression coverage.
• Responsible human: @youngsofun
• The responsible human has read every line of this diff and can explain each change

---

This change is 

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9f0a01fcaa

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".