compute_volume_residual_fields / boundary sibling still strip Dirichlet values (the #407 fix was applied to one of three duplicated variants)

[lmoresi]

Follow-up from the #407 fix review (2026-07-24): _assemble_volume_reaction got the DMPlexInsertBoundaryValues insert, but the duplicated round-trip in compute_volume_residual_fields and its boundary sibling (petsc_generic_snes_solvers.pyx ~7557) did not — their docstrings advertise the same CBF use-case and they still return garbage on g != 0 walls. No in-repo callers today.

This trio is already flagged as duplication in the readability review (READ-27): the right fix is probably to de-duplicate onto one insert-correct core rather than patch three copies.

Underworld development team with AI support from Claude Code

[lmoresi]

Fixed by PR #456, merged to development on 2026-07-28. (Manual close — PRs target development, so auto-close doesn't fire.)