Make PST work for different smoothing length factors

[efaulhaber]

On main, PST doesn't work for small smoothing length factors. The following examples are all with the Wendland C2 kernel.

	h = 1.5 * Δx	h = 2.5 * Δx
Δx = 0.02, Δt = 5e-3

This PR makes it work for a wide range of time steps, smoothing length factors, and resolutions:

	h = 1.5 * Δx	h = 2.5 * Δx
Δx = 0.02, Δt = 1e-3
Δx = 0.02, Δt = 5e-3
Δx = 0.01, Δt = 5e-3
Δx = 0.05, Δt = 1e-2

[Copilot]

Pull Request Overview

This PR refactors the particle shifting callback to correctly handle a range of smoothing length factors by introducing a dedicated dx variable and updating the shift magnitude calculation.

• Extracted particle_spacing(system, 1) into dx and reused it for kernel evaluation.
• Replaced the original linear-in-h formula with a quadratic-in-h and linear-in-dx scaling for delta_r_, adding comments to explain the change.

Comments suppressed due to low confidence (2)

src/callbacks/particle_shifting.jl:108

• [nitpick] Referencing an internal PR link in comments can be brittle; consider citing the original paper (Sun et al. 2017, p.29) or documenting the rationale in the external changelog instead.

# See https://github.com/trixi-framework/TrixiParticles.jl/pull/834.

src/callbacks/particle_shifting.jl:109

• The new shifting formula significantly alters particle behavior; consider adding unit or integration tests that verify correct behavior across small and large smoothing length factors.

delta_r_ = -dt * v_max * (2 * h)^2 / (2 * dx) * (1 + R * (kernel / Wdx)^n) *

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes missing coverage. Please review.

Project coverage is 70.45%. Comparing base (ac2acd2) to head (0909b81).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/callbacks/particle_shifting.jl	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #834      +/-   ##
==========================================
- Coverage   70.46%   70.45%   -0.02%     
==========================================
  Files         106      106              
  Lines        6894     6895       +1     
==========================================
  Hits         4858     4858              
- Misses       2036     2037       +1     

Flag	Coverage Δ
unit	70.45% <0.00%> (-0.02%)	⬇️

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.

[efaulhaber]

Please verify that this is also correct for 3D cases.

This is a slice of the same vortex street simulation in 3D, where the cylinder is now a full sphere.
Top is 1.5Δx, bottom is 2.5Δx. Both with this branch.

On main:

So I guess this is fine in 3D. We will probably only be using consistent shifting in the future anyway.

[svchb]

/run-gpu-tests