[Backport 0.18] fix(hgraph): tune from available fp32 codes

[jac0626]

Summary

Backport #2466 to the 0.18 branch.

• Let HGraph Tune select an available lossless source from raw vectors, precise FP32 codes,
  or base FP32 codes instead of relying only on store_raw_vector.
• Require the selected codes to cover the current internal ID range.
• Rebuild precise codes whose payload was omitted by ignore_reorder serialization.
• Keep Tune state publication atomic and add source-selection and serialization regression
  coverage.

Fixes: #2465

Backport adaptation

• Ported the implementation to the 0.18 single-file HGraph layout.
• Omitted reorder_source handling because that API does not exist in 0.18.
• Omitted the ForceRemove regression because RemoveMode::FORCE_REMOVE is not available in
  0.18.

Validation

/opt/homebrew/opt/llvm@15/bin/clang-format --dry-run --Werror \
  src/algorithm/hgraph.cpp tests/test_hgraph.cpp
git diff --check upstream/0.18..HEAD

The 0.18 test binary was not run locally: its legacy build configuration is incompatible with
the current macOS toolchain before reaching the changed code. CI is expected to provide the
target-platform validation.

Compatibility

• API/ABI impact: none.

[vsag-bot]

/label S-ready-to-merge
/waiting-on maintainer
/request-review @jiaweizone
/request-review @inabao

[vsag-bot]

Automated pull request review completed.

Review effort: medium (186 changed lines across 2 files).

Submitted 1 inline comment.
Review: #2476 (review)

[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+)

[gemini-code-assist]

Code Review

This pull request refactors the HGraph::Tune method to optimize how source vector data is selected and decoded during tuning, introducing helper lambdas and checking for active IDs. It also moves updates to create_param_ptr_ inside the global_mutex_ lock and adds comprehensive unit tests for various tuning scenarios. The reviewer identified two critical concurrency issues: a potential data race when reading immutable_ under add_mutex_ while it is modified under global_mutex_, and another data race when modifying create_param_ptr_ fields while concurrent readers like ExportModel and Fork access them without locking.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[vsag-bot]

Automated inline review completed.

Review effort: medium (185 changed lines across 2 files).
No actionable inline findings were found.

Reviewed commit 89f854f.

[vsag-bot]

Automated inline review completed.

Review effort: medium (186 changed lines across 2 files).
No actionable inline findings were found.

Reviewed commit 504828c.

[vsag-bot]

Automated inline review completed.

Review effort: medium (186 changed lines across 2 files).
No actionable inline findings were found.

Reviewed commit f7ef7b9.

[vsag-bot]

Automated inline review completed.

Review effort: medium (186 changed lines across 2 files).
Submitted 1 inline comment.

Reviewed commit 130f284.

[codecov]

Codecov Report

❌ Patch coverage is 86.95652% with 6 lines in your changes missing coverage. Please review.

@@            Coverage Diff             @@
##             0.18    #2476      +/-   ##
==========================================
+ Coverage   91.29%   91.33%   +0.03%     
==========================================
  Files         330      330              
  Lines       19657    19678      +21     
==========================================
+ Hits        17946    17972      +26     
+ Misses       1711     1706       -5     

Flag	Coverage Δ
cpp	91.33% <86.95%> (+0.03%)	⬆️

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

Components	Coverage Δ
common	85.94% <ø> (ø)
datacell	93.25% <ø> (+0.46%)	⬆️
index	91.51% <86.95%> (-0.04%)	⬇️
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 feb5e4a...130f284. 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.

[wxyucs]

lgtm