PR Review for #62

[rsenne]

This PR is for fixing things I didn't address because i accidentally merged #63 and #62.

[Copilot]

Pull request overview

This PR completes and stabilizes the work from #62 by allowing DensityModel derivative slots to accept AD backends, preparing those backends once per chain (cached in sampler state), and adjusting DEER/HVP logic and related extensions/docs/tests accordingly.

Changes:

• Add AD-backend-or-callable support for DensityModel derivative slots and introduce PreppedDensityModel with per-chain preparation reuse.
• Rework DEER HVP strategy/normalization and add second-order HVP support that routes through DI’s SecondOrder where appropriate.
• Expand test coverage and update extensions + documentation to reflect the new slot semantics and optional ParallelMALASampler(…; backend=…).

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
test/test-Turing-Integration.jl	Adds integration tests for SecondOrder HVP on Turing/DynamicPPL models and for batched-slot routing to the batched DEER path.
test/test-HVP-Strategy.jl	Adds regression tests for backend normalization and SecondOrder normalization preserving DI-resolved composition.
test/test-GPU-DEER.jl	Adjusts GPU backend collection (notably Enzyme) for DEER GPU tests.
test/test-DEER-Interface.jl	Updates DEER interface tests to use _prepare_model and pass prepped models to DEER recursion builders.
test/test-ADType-Slots.jl	New comprehensive test suite for AD-backend slots, preparation semantics, and state/model reuse behavior.
src/ParallelMCMC.jl	Introduces module-level DI aliasing and imports needed for AD-backend slot support.
src/interface.jl	Implements PreppedDensityModel, slot validation, backend resolution, per-chain preparation caching in states, and makes ParallelMALASampler.backend optional.
src/DEER/DEER.jl	Refactors backend normalization and adds DI SecondOrder HVP helpers; adjusts HVP fallback routing and preparation paths.
ext/LogDensityProblemsExt.jl	Forwards batched slots and documents SecondOrder requirement for AD HVP on prepared-gradient LDP models.
ext/EnzymeExt.jl	Updates normalization hook to only fill function_annotation=Enzyme.Const without forcing a mode.
ext/DynamicPPLExt.jl	Forwards batched slots and documents SecondOrder requirement for AD HVP on DynamicPPL-prepared gradients.
docs/src/90-contributing.md	Fixes list formatting in contributing guide.
docs/src/15-gpu.md	Updates GPU guidance around log-density-only models, second-order limitations, and optional sampler backend.
docs/src/10-getting-started.md	Adds a new section documenting AD backends in derivative slots and clarifying HVP semantics.
CHANGELOG.md	Adds a detailed changelog entry describing the new slot semantics, caching behavior, and backend normalization changes.

Suppressed comments (1)

src/interface.jl:710

• Changing _build_mala_deer_rec to require a PreppedDensityModel breaks existing internal callers that still pass a DensityModel with backend= (e.g. benchmark scripts under benchmarks/ParallelMCMCBenchmarks/scripts/*), causing a MethodError. Consider keeping a thin wrapper method that accepts DensityModel + backend and forwards to the prepped-model method.

function _build_mala_deer_rec(
    model::PreppedDensityModel,
    ε::Real,
    tape::Vector{<:MALATapeElement},
    x0_like::AbstractVector;

---

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

[codecov]

Codecov Report

❌ Patch coverage is 87.50000% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 92.05%. Comparing base (b43b16e) to head (ad1f64c).

Files with missing lines	Patch %	Lines
src/DEER/DEER.jl	87.50%	1 Missing ⚠️

❌ Your patch check has failed because the patch coverage (87.50%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #65      +/-   ##
==========================================
- Coverage   92.07%   92.05%   -0.03%     
==========================================
  Files           7        7              
  Lines        1173     1170       -3     
==========================================
- Hits         1080     1077       -3     
  Misses         93       93              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

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

[rsenne]

@gdalle this is to address your concerns on #62 that i merged by mistake