[minmax] Refactor full query

[arkrishn94]

Refactor full precision queries for minmax to use Slice and its variants.

[Copilot]

Pull request overview

Refactors MinMax full-precision query handling to use the shared meta::slice abstractions (Slice/PolySlice and borrowed variants), reducing bespoke query container code and aligning query handling with the rest of the quantization metadata-slice patterns.

Changes:

• Replaces the bespoke FullQuery struct in MinMax with slice::PolySlice and introduces FullQueryRef / FullQueryMut.
• Updates MinMax distance function impls and the MinMax quantizer to operate on FullQueryRef/FullQueryMut and Slice accessors (vector(), meta()).
• Adds Reborrow/ReborrowMut support for owning PolySlice and updates benchmark wiring accordingly.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
diskann-quantization/src/minmax/vectors.rs	Replaces FullQuery struct with PolySlice aliases; updates distance functions and tests to use FullQueryRef and slice accessors.
diskann-quantization/src/minmax/quantizer.rs	Updates full-precision query compression to take FullQueryMut and write via vector_mut() / meta_mut().
diskann-quantization/src/minmax/multi/meta.rs	Formatting-only change (blank line).
diskann-quantization/src/minmax/mod.rs	Re-exports new FullQueryRef/FullQueryMut alongside FullQuery.
diskann-quantization/src/meta/slice.rs	Adds Reborrow / ReborrowMut impls for owning PolySlice to produce SliceRef/SliceMut.
diskann-benchmark/src/backend/exhaustive/minmax.rs	Updates exhaustive benchmark to use FullQueryRef and allocate FullQuery via new_in.

Comments suppressed due to low confidence (1)

diskann-quantization/src/minmax/vectors.rs:168

• The doc comment for FullQueryMeta describes the inner product as between X = ax * X' + bx and Y, but the implementation for full-precision queries applies a/b from y.meta() (the quantized vector), not from the query. Please update the formula/variable names to match the code path (full-precision query vs quantized data vector) to avoid confusion.

/// The inner product between `X = ax * X' + bx` and `Y` for d-dimensional
/// vectors X and Y is:
/// ```math
/// <X, Y> = <ax * X' + bx, Y>
///        = ax * <X', Y> + bx * sum(Y).

---

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

[arkrishn94]

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

[codecov-commenter]

Codecov Report

❌ Patch coverage is 94.87179% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.98%. Comparing base (87a0757) to head (1071f89).
⚠️ Report is 4 commits behind head on main.

Files with missing lines	Patch %	Lines
diskann-quantization/src/minmax/quantizer.rs	83.33%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #770   +/-   ##
=======================================
  Coverage   88.98%   88.98%           
=======================================
  Files         428      428           
  Lines       78365    78415   +50     
=======================================
+ Hits        69730    69781   +51     
+ Misses       8635     8634    -1     

Flag	Coverage Δ
miri	88.98% <94.87%> (+<0.01%)	⬆️
unittests	88.98% <94.87%> (+<0.01%)	⬆️

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

Files with missing lines	Coverage Δ
diskann-benchmark/src/backend/exhaustive/minmax.rs	100.00% <ø> (ø)
diskann-quantization/src/meta/slice.rs	99.50% <100.00%> (+0.03%)	⬆️
diskann-quantization/src/minmax/vectors.rs	94.77% <100.00%> (+0.69%)	⬆️
diskann-quantization/src/minmax/quantizer.rs	91.55% <83.33%> (-0.63%)	⬇️

... and 21 files with indirect coverage changes

🚀 New features to boost your workflow:

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