Adding tests for diverse_search, diverse_priority_queue, and a small transient suite.

[JordanMaples]

I had Copilot take a look at some of the test coverage numbers and write up some unit tests to address some
gaps.

• diverse_search was only gaining test coverage in diskann-providers, so unit tests were added directly in
  diskann. The test in providers will remain as it is an integration test and requires tight coupling with inmem.
• Added some transient error tests that were gaps highlighted following the implementation of the flaky tests
  added in Migrate test_flaky_consolidate from diskann_providers to diskann #942.
• Added some extra validation in diverse_priority_queue.

[Copilot]

Pull request overview

This PR expands unit test coverage in diskann for experimental diverse search and for transient (non-critical) error handling paths, and adjusts the in-memory test provider strategy to optionally inject transient accessor failures during search.

Changes:

• Added/expanded tests for DiverseNeighborQueue post-processing and eviction bookkeeping.
• Added a new transient error test suite covering search beam-expansion and insert-time pruning tolerance.
• Added crate-level diverse search tests (behind experimental_diversity_search) and enabled flaky accessors in the test provider strategy.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
diskann/src/neighbor/diverse_priority_queue.rs	Adds stronger assertions and new post_process edge-case tests for the diverse priority queue.
diskann/src/graph/test/provider.rs	Enables transient/flaky accessor injection via Strategy::with_transient for search/prune accessors.
diskann/src/graph/test/cases/transient_errors.rs	New tests ensuring transient accessor failures are tolerated during search and insert pruning.
diskann/src/graph/test/cases/mod.rs	Registers the new test modules (diverse_search behind feature flag; transient_errors always).
diskann/src/graph/test/cases/diverse_search.rs	New crate-level tests exercising diverse search behavior and parameter accessors.

---

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

[Copilot]

diverse_search_div_l_zero_panics is currently driven by diverse_results_k = 0 (run_diverse_search(..., 10, 10, 0, ...)), not by the documented rounding case (diverse_results_k * l / k == 0 with diverse_results_k > 0). This makes the test misleading and effectively encodes “invalid input causes an internal underflow panic”. Prefer adding/using input validation (e.g., reject diverse_results_k == 0 with a clear error) and updating this test to assert the validation behavior rather than #[should_panic] on an internal implementation detail.

Suggested change

	// ─── diverse_results_l boundary: div_l = 0 panics ────────────────────────────
	#[test]
	#[should_panic]
	fn diverse_search_div_l_zero_panics() {
	// When diverse_results_k * l_value / k_value rounds to 0, local queues
	// have capacity 0. The first insert with an attribute triggers Case 2
	// (local queue "full" at size 0 == capacity 0), which indexes at
	// diverse_results_l - 1 = usize::MAX. This is an unchecked edge case.
	let (index, attr) = setup_diverse_grid(Grid::Two, 4, 3);
	let query = vec![0.0f32; 2];
	// k=10, l=5, div_k=1 → diverse_results_l = 1 * 5 / 10 = 0
	let _result = run_diverse_search(&index, &query, 10, 10, 0, attr);
	// ─── Input validation: diverse_results_k must be non-zero ────────────────────
	#[test]
	fn diverse_search_rejects_zero_diverse_results_k() {
	let (index, attr) = setup_diverse_grid(Grid::Two, 4, 3);
	let query = vec![0.0f32; 2];
	let panic = std::panic::catch_unwind(|| {
	let _result = run_diverse_search(&index, &query, 10, 10, 0, attr);
	})
	.expect_err("diverse_results_k == 0 should be rejected by input validation");
	let panic_message = if let Some(message) = panic.downcast_ref::<&str>() {
	*message
	} else if let Some(message) = panic.downcast_ref::<String>() {
	message.as_str()
	} else {
	panic!("validation panic should use a string message");
	};
	assert!(
	panic_message.contains("diverse_results_k")
	&& (panic_message.contains("non-zero") || panic_message.contains("zero")),
	"unexpected panic message for invalid diverse_results_k: {panic_message}",
	);

[JordanMaples]

This one is a pretty valid comment. it seems to me that diverse_results_k should be a NonZeroUsize, it's not really clear to me why we'd allow 0 as a value.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 97.46835% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.44%. Comparing base (56e3b7f) to head (912b605).

Files with missing lines	Patch %	Lines
diskann/src/graph/test/cases/transient_errors.rs	97.36%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #972      +/-   ##
==========================================
- Coverage   90.58%   89.44%   -1.15%     
==========================================
  Files         449      450       +1     
  Lines       83779    83857      +78     
==========================================
- Hits        75891    75002     -889     
- Misses       7888     8855     +967     

Flag	Coverage Δ
miri	89.44% <97.46%> (-1.15%)	⬇️
unittests	89.28% <97.46%> (-1.27%)	⬇️

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

Files with missing lines	Coverage Δ
diskann/src/graph/test/provider.rs	94.41% <100.00%> (+0.01%)	⬆️
diskann/src/graph/test/cases/transient_errors.rs	97.36% <97.36%> (ø)

... and 38 files with indirect coverage changes

🚀 New features to boost your workflow:

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