Fix benchmark ground truth for disk_index_siftsmall 10-query workload

[partychen]

• Does this PR have a descriptive title that could go in our release notes?
• Does this PR add any new dependencies?
• Does this PR modify any existing APIs?
• Is the change to the API backwards compatible?
• Should this result in any changes to our documentation, either updating existing docs or adding new ones?

Reference Issues/PRs

Fixes #1172.

What does this implement/fix? Briefly explain your changes.

This PR adds an exact brute-force ground-truth file for benchmark examples using
the disk_index_siftsmall_learn_256pts_data.fbin and
disk_index_sample_query_10pts.fbin workload.

• Added
  test_data/disk_index_search/disk_index_10pts_idx_uint32_exact_ground_truth.bin,
  generated using compute_groundtruth.
• Updated the applicable benchmark example configurations to use the new exact
  ground-truth file.
• Added no dependencies and made no API changes.

This fixes the recall mismatch where flat-search could report recall below
1.0 because the benchmark configurations previously referenced results that
were not exact brute-force ground truth.

Any other comments?

Validated with:

• cargo test --locked --workspace
• All benchmark example JSON files parse successfully.

[Copilot]

Pull request overview

Updates the ground-truth fixture used by the disk_index_siftsmall_learn_256pts_data.fbin + disk_index_sample_query_10pts.fbin 10-query benchmark workload so that exhaustive/flat-search benchmarks compare against brute-force KNN results (fixing the recall < 1.0 mismatch described in #1172).

Changes:

• Replaced the Git LFS object for disk_index_10pts_idx_uint32_truth_search_res.bin with newly generated brute-force ground truth.
• Increased the ground-truth payload size (408 → 808 bytes), consistent with compute_groundtruth’s on-disk format of Metadata + ids + distances for 10 queries × K=10.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[harsha-simhadri]

Thanks! Are there any other workflows using the older file? Asking to check if it might break any workflow.

[partychen]

Thanks! Are there any other workflows using the older file? Asking to check if it might break any workflow.

Thanks for calling this out. I checked all references to the existing file.

The disk-provider regression tests still use the older
disk_index_10pts_idx_uint32_truth_search_res.bin file and expect its contents
to remain unchanged.

The benchmark examples were the other consumers. They now reference the new
disk_index_10pts_idx_uint32_exact_ground_truth.bin file, which contains the
brute-force KNN ground truth.

So the existing file remains unchanged for workflows that depend on its current
results, while benchmark recall calculations use the new exact ground-truth
file. I also ran cargo test --locked --workspace, and all tests passed.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.16%. Comparing base (62e201a) to head (0cfc89f).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1261      +/-   ##
==========================================
- Coverage   90.16%   90.16%   -0.01%     
==========================================
  Files         509      509              
  Lines       97234    97238       +4     
==========================================
+ Hits        87672    87675       +3     
- Misses       9562     9563       +1     

Flag	Coverage Δ
miri	90.16% <ø> (-0.01%)	⬇️
unittests	89.83% <ø> (-0.01%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.
see 5 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.