fix(ddt): units-active semi-Lagrangian trace-back (#267)

[lmoresi]

Summary

Fixes #267 — SemiLagrangian.update_pre_solve crashed for any unit-bearing model in the time loop:

ValueError: Cannot subtract arrays with incompatible units: 'meter' and 'meter / second'

Root cause

The SL trace-back is performed in the mesh's non-dimensional (DM) coordinate space — evaluate/global_evaluate treat plain arrays as DM coords, and DM point-location uses DM values (0..L_model, not dimensional metres). The has_units branch instead kept dimensional coords (mesh.X.coords ≈ metres, ~1e9) and velocity (m/s) and left dt a unitless float, so coords[m] − v[m/s]·dt[dimensionless] raised — and even patched, dimensional coords fed to a DM that locates in 0..1000 space would mislocate.

Fix

Route the whole trace-back through ND/DM space regardless of units:

• departure point from psi_star[i].coords_nd (== .coords for a non-units model; the ND reduction of dimensional coords when units are active),
• velocity non-dimensionalised,
• dt non-dimensional.

Both RK2 legs (node-velocity and midpoint-velocity) corrected. The non-units path is unchanged — it already used exactly this ND logic; the fix just removes the broken parallel dimensional branch.

Verification

• Units-active AdvDiffusionSLCN now runs and tracks the equivalent non-dimensional run to ~1e-3 (the small residual is the constitutive diffusivity scaling under units — a separate concern, not the trace-back; advection-magnitude-independent).
• Non-units SL regression unchanged: test_0855 (oldframe SL), test_0610 (NS SLCN), test_1054, test_1100 pass.
• test_1110_advDiffAnnulus fails identically with and without this change — a pre-existing PCSetUp_MG/DMCreateInjection multigrid issue, unrelated.
• New regression: tests/test_1056_units_slcn_traceback.py (runs + nondim-equivalence).

Unblocks the time-loop half of #263 (Tutorial_Thermal_Convection_Units).

Underworld development team with AI support from Claude Code

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[lmoresi]

Self-review: routes the SL trace-back through ND/DM space (the non-units path's existing logic); removes the broken dimensional branch. Repro fixed, regression test passes (2/2), non-units SL suite unchanged (the one annulus failure is a pre-existing MG issue, identical with/without). Merging for the release.