fix(sql): handle correlated subqueries over union

[sundy-li]

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

Summary

• preserve correlated-column mappings across decorrelated UNION / UNION ALL branches
• add replay and service regressions for correlated EXISTS subqueries over set operations

Changes

• compute branch-level need_cross_join in flatten_sub_union_all whenever a branch does not cover the full correlated key set
• replace unwrap()-based correlated mapping lookups with get_derived() so missing mappings surface as errors instead of panics
• add shared optimizer replay cases for the reported UNION and UNION ALL repros
• add service-side regression tests for the same two query shapes

Tests

• Unit Test

• Logic Test

• Benchmark Test

• No Test - Pair with the reviewer to explain why

• TEST_SUBDIR=regressions cargo test --package databend-common-sql --test it -- planner::test_lite_replay_service_optimizer_cases --exact --nocapture

• cargo fmt --all

Type of change

• Bug fix

Fixes #19574

---

This change is 

[github-actions]

🤖 CI Job Analysis

Workflow: 23490233034

📊 Summary

• Total Jobs: 86
• Failed Jobs: 1
• Retryable: 0
• Code Issues: 1

❌ NO RETRY NEEDED

All failures appear to be code/test issues requiring manual fixes.

🔍 Job Details

• ❌ linux / test_unit: Not retryable (Code/Test)

---

🤖 About

Automated analysis using job annotations to distinguish infrastructure issues (auto-retried) from code/test issues (manual fixes needed).