Our test_cancel_optimization test is flaky. It fails on high CPU systems.

$ QDRANT_NUM_CPUS=24 cargo test -p collection -- test_cancel_optimization

---- tests::test_cancel_optimization stdout ----
actual_optimization_duration = 114 ms

thread 'tests::test_cancel_optimization' panicked at lib/collection/src/tests/mod.rs:210:9:
assertion `left == right` failed
  left: 2
 right: 3

The reason is asymmetric utilization of IO and CPU budget (implemented in #6015). On 24 CPUs, the previous implementation allowed 2 concurrent IO tasks while it allowed 3 concurrent CPU tasks.

This adjusts the permit selection to make it symmetric.

If we don't like this, the alternative would be to adjust the test. Change this to:

// We skip optimizations that use less than half of the preferred CPU budget
let expected_optimization_count = {
    let cpus = common::cpu::get_cpu_budget(0);
    let hnsw_threads = num_rayon_threads(0);
    (cpus / hnsw_threads).clamp(1, 3)
};

All Submissions:

• Contributions should target the dev branch. Did you create your branch from dev?
• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same update/change?

New Feature Submissions:

1. Does your submission pass tests?
2. Have you formatted your code locally using cargo +nightly fmt --all command prior to submission?
3. Have you checked your code using cargo clippy --all --all-features command?