fix: preserve sorts for wrapping integer negation

[haohuaijin]

Which issue does this PR close?

Closes #24683.
Related to #25407.

Rationale for this change

An ordered signed integer column does not necessarily remain ordered after negation: the minimum value wraps to itself. Treating negation as always reversing order can remove a required sort and return incorrectly ordered rows. Negating inferred interval endpoints can also raise an overflow error even though array evaluation uses wrapping negation.

What changes are included in this PR?

• Stop deriving reversed ordering when the input range may contain the signed minimum, while preserving existing Singleton properties and reversal for known safe ranges.
• Use unbounded intervals for potentially wrapping signed integer negation in forward and reverse constraint propagation, including singleton MIN ranges.
• Add regression cases to the existing order.slt and update the expected plan for negation of an ordered BIGINT expression.

Runtime scalar and array negation semantics are unchanged. This conservative fix can retain additional sorts when safe bounds are unavailable, including after widening casts; improving cast range propagation is left to separate work.

What is the testing strategy for this PR?

• A table-driven unit test covers all four signed integer widths, unbounded and MIN ranges, Singleton properties, safe reversal, and unknown input properties.
• Four cases in order.slt cover the original incorrect ordering, MIN filter propagation, double negation on an unbounded stream, and ordering MIN values mixed with NULL.
• The original ordering query and MIN filter reproduced failures before the fix.
• Passed the 9 negation unit tests and order.slt, plus formatting, full Clippy, and uv run ./dev/rust_lint.sh checks.

Are there any user-facing changes?

Queries involving signed minimum negation retain necessary sorts and no longer fail solely because an inferred negation endpoint overflows. Some range refinement and sort elimination become more conservative. No public API changes.

[haohuaijin]

As a follow-up to #25220, exact column statistics could prove that a + b cannot reach INT_MIN, allowing us to safely reverse the declared ordering and eliminate this SortExec.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.65%. Comparing base (76f9fde) to head (3e4cafb).

Additional details and impacted files

@@           Coverage Diff            @@
##             main   #25568    +/-   ##
========================================
  Coverage   82.65%   82.65%            
========================================
  Files        1147     1147            
  Lines      446087   446205   +118     
  Branches   446087   446205   +118     
========================================
+ Hits       368710   368821   +111     
- Misses      54990    54994     +4     
- Partials    22387    22390     +3     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[haohuaijin]

Hi @kosiew, could you help review this follow-up to #25234 for signed-minimum negation? Thanks!

[kosiew]

@haohuaijin,

Thanks for working on this. The wrapping signed-integer negation handling looks correct, and the new tests cover the main ordering and interval-propagation cases well. I have one non-blocking test suggestion below.

[haohuaijin]

Thanks @kosiew , i added the test case in 5e270ae