fix: HGraph dirty-vector search returns fewer results than top-k

[vsag-bot]

Summary

HGraph search now traverses NaN/Inf bridge nodes deterministically while excluding NaN distances from results, preventing incomplete top-k responses.

Root Cause

Non-finite distances were inserted directly into candidate heaps using negation. NaN violates heap ordering and candidate admission comparisons, causing traversal to stop before enough finite neighbors were collected.

Changes

• Added shared result-eligibility and traversal-priority helpers for non-finite distances.
• Updated basic, iterative, and parallel graph search paths to traverse non-finite bridge nodes without returning NaN results.
• Added focused BasicSearcher and ParallelSearcher regressions covering NaN and overflow-to-Inf bridges.

Validation

• cmake --build build-ai-fix --target unittests functests --parallel 4: passed - Debug unit and functional test binaries built successfully.
• build-ai-fix/tests/unittests '*Searcher traverses through a non-finite-distance bridge' --reporter compact: passed - 12 assertions passed across both focused regressions; the BasicSearcher regression failed before the fix.
• build-ai-fix/tests/unittests '[ut][BasicSearcher]' --reporter compact && build-ai-fix/tests/unittests '[ut][ParallelSearcher]' --reporter compact: passed - 243422 assertions passed across the relevant searcher suites.
• for run in 1 2 3 4 5; do build-ai-fix/tests/functests '(PR) HGraph Search with Dirty Vector' --reporter compact --rng-seed "$run"; done: passed - All five runs passed with 283 assertions each.
• git diff --check: passed - No whitespace errors found.

Residual Risks

• The sanitizer workflow was not reproduced locally; validation used a debug build because CMake 3.22 could not directly consume the provided local archive fallback lists.

Base branch: 0.18

Fixes #2732

Prepared by vsag-bot after automatic effort ordering and AI repairability preflight.

[vsag-bot]

/label status/waiting-for-review
/waiting-on reviewer
/request-review @inabao
/request-review @ShawnShawnYou
/request-review @LHT129

[vsag-bot]

AI PR state: status/ai-working
Current owner: @vsag-bot
Reason: checks concluded failure
Head: fd7e661

Automatic merging is disabled. A maintainer performs the final merge.

[mergify]

Merge Protections

🟢 Merge protection satisfied — ready to merge.

Show 1 satisfied protection

🟢 Require linked issue for feature/bug PRs

• body~=(?im)(?:^|[\s\-\*])(?:close[sd]?|fix(?:e[sd])?|resolve[sd]?)\s*:?\s+(?:#\d+|[\w.\-]+/[\w.\-]+#\d+|https?://github\.com/[\w.\-]+/[\w.\-]+/issues/\d+)

[vsag-bot]

AI PR CI recovery

Item	Value
Status	branch-updated
Head	c51b1be
Workflow	Asan Build & Test Parallel
Classification	change_related
Confidence	99%
Action	Pushed corrective commit fd7e661; CI will run for the new head

Fixed the ParallelSearcher worker lifetime race reported by ASan.

Reasoning:
Both architectures failed in the newly added ParallelSearcher regression. Worker tasks captured stack-owned queues, but Search returned without waiting for workers to consume shutdown markers. Retaining and waiting on their futures ensures all workers exit before local state is destroyed.

[codecov]

Codecov Report

❌ Patch coverage is 90.69767% with 4 lines in your changes missing coverage. Please review.

@@            Coverage Diff             @@
##             0.18    #2734      +/-   ##
==========================================
- Coverage   91.39%   91.30%   -0.10%     
==========================================
  Files         331      332       +1     
  Lines       19714    19733      +19     
==========================================
- Hits        18018    18017       -1     
- Misses       1696     1716      +20     

Flag	Coverage Δ
cpp	91.30% <90.69%> (-0.10%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Components	Coverage Δ
common	86.00% <ø> (ø)
datacell	91.66% <ø> (-1.30%)	⬇️
index	91.72% <ø> (+0.01%)	⬆️
simd	100.00% <ø> (ø)

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 89188a9...fd7e661. Read the comment docs.

🚀 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.