Make continuity_equation! and density_diffusion! work on Ref values

[efaulhaber]

This PR makes continuity_equation! and density_diffusion! work on Ref values instead of directly writing to dv. Together with #1116, it makes it possible to accumulate drho_particle over all neighbors and only once write it to dv.
It is part of #1131.

[codecov]

Codecov Report

❌ Patch coverage is 94.36620% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 89.06%. Comparing base (419d5b9) to head (5f36a06).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/schemes/fluid/shifting_techniques.jl	63.63%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1130      +/-   ##
==========================================
+ Coverage   89.00%   89.06%   +0.06%     
==========================================
  Files         128      128              
  Lines        9838     9868      +30     
==========================================
+ Hits         8756     8789      +33     
+ Misses       1082     1079       -3     

Flag	Coverage Δ
total	89.06% <94.36%> (+0.06%)	⬆️
unit	67.18% <76.05%> (+0.05%)	⬆️

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

☔ View full report in Codecov by Sentry.
📢 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.

[Copilot]

Pull request overview

This PR refactors density-rate updates so continuity_equation! and density_diffusion! can accumulate into a Ref (e.g., drho_particle) rather than writing directly into dv, enabling a later optimization to write dv once per particle (as described in #1116 / #1131).

Changes:

• Update fluid WCSPH and EDAC RHS interaction loops to pass v_a/v_b and accumulate drho_particle::Ref via continuity_equation!, then write via write_drho_particle!.
• Refactor density_diffusion! to accumulate into drho_particle::Ref and adjust call sites accordingly.
• Adjust shifting-technique helpers to operate on/return a transformed v_diff value.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/schemes/structure/structure.jl	Switch structure–fluid continuity update to accumulate into Ref and add a write_drho_particle! writeback path.
src/schemes/fluid/weakly_compressible_sph/rhs.jl	Accumulate density-rate in Ref and write once via write_drho_particle!.
src/schemes/fluid/weakly_compressible_sph/density_diffusion.jl	Make density diffusion accumulate into Ref instead of writing to dv.
src/schemes/fluid/shifting_techniques.jl	Update shifting continuity helpers to accept/return v_diff directly.
src/schemes/fluid/fluid.jl	Refactor continuity_equation! to write into Ref and introduce write_drho_particle! dispatch.
src/schemes/fluid/entropically_damped_sph/rhs.jl	Use Ref accumulation for continuity equation and defer writeback.
src/schemes/boundary/wall_boundary/rhs.jl	Use Ref accumulation for boundary dummy-particle continuity update.
src/schemes/boundary/open_boundary/rhs.jl	Update density diffusion call to use Ref accumulation (continuity term remains direct).

---

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

[svchb]

I found one blocker in the current head.

Blocker: open-boundary continuity path lost its dispatch target

In src/schemes/boundary/open_boundary/rhs.jl, the PR now calls:

continuity_equation!(drho_particle, ContinuityDensity(),
                     particle_system, neighbor_system, ...)

where particle_system is an OpenBoundarySystem{<:BoundaryModelDynamicalPressureZhang}.

But after the refactor, the relevant overload in src/schemes/fluid/fluid.jl is only defined for

particle_system::AbstractFluidSystem

and OpenBoundarySystem is not an AbstractFluidSystem; it is an AbstractSystem{NDIMS}.

Before this PR, there was still a generic helper

continuity_equation!(dv, particle_system, neighbor_system, ...)

with no AbstractFluidSystem restriction, and the open-boundary code reused that path. This PR removes that generic path by folding the logic into the typed overload, so this looks like a regression that should surface as a MethodError once the open-boundary interaction is exercised.

Why I think this is the intended fix direction

The helpers used inside the continuity update already have open-boundary support: OpenBoundarySystem defines density_diffusion and shifting_technique, and density_diffusion! already accepts OpenBoundarySystem{<:BoundaryModelDynamicalPressureZhang}. So the missing piece is the continuity_equation! dispatch, not the underlying math.

Suggested fix

Either:

1. Restore the generic helper that computes the continuity contribution for any system supporting the required accessors, then let both fluid and open-boundary code call it, or
2. Add a dedicated overload for OpenBoundarySystem{<:BoundaryModelDynamicalPressureZhang} mirroring the previous generic behavior.

What looks fine now

The earlier Copilot concern about write_drho_particle! for structure systems seems already addressed: structure.jl now has a generic no-op write_drho_particle!(dv, ::AbstractSystem, ...), so that specific MethodError no longer looks live.

So my review would be: request changes for the open-boundary continuity dispatch regression.

[efaulhaber]

/run-gpu-tests