Fixing flaky test_compute_relative_contrast_with_random_data

[arrayka]

What

• Use a seeded pseudo-random generator in the test

Why

• To fix flaky test_compute_relative_contrast_with_random_data() test

Example of the failure: https://github.com/microsoft/DiskANN/actions/runs/21272293270/job/61224767878?pr=697

    test utils::relative_contrast::relative_contrast_tests::test_compute_relative_contrast_with_random_data ... FAILED

    failures:

    failures:
        utils::relative_contrast::relative_contrast_tests::test_compute_relative_contrast_with_random_data

    test result: FAILED. 0 passed; 1 failed; 0 ignored; 0 measured; 14 filtered out; finished in 0.50s
    
  stderr ───

    thread 'utils::relative_contrast::relative_contrast_tests::test_compute_relative_contrast_with_random_data' panicked at diskann-tools/src/utils/relative_contrast.rs:192:9:
    Mean relative contrast is out of range: 1.2039199

[arrayka]

@microsoft-github-policy-service agree company="Microsoft"

[hildebrandmw]

Thanks Alex - we should be sure to port this (and the other changes made to this repo) internally to reduce the merge headache when we finalize the transition.

[arrayka]

Thanks Alex - we should be sure to port this (and the other changes made to this repo) internally to reduce the merge headache when we finalize the transition.

@hildebrandmw, ack