Fix three pre-existing multi-rank broadcast defects; deregister dead parameters

[sbryngelson]

Stacked on #1551 and the executable-dedup PR — this branch includes both; its own content is the top 3 commits. It exists separately because, unlike its parents, it intentionally changes multi-rank behavior (it fixes bugs).

Summary

While converting the MPI broadcast lists to registry-driven generation (parent PR), a member-by-member audit of every namelist parameter's broadcast/consumption pattern turned up three pre-existing defects on master. All three predate this work.

1. bc_x/y/z%{vb1..3, ve1..3} were never broadcast in pre_process/post_process but are consumed on all ranks by s_slip_wall/s_no_slip_wall in m_boundary_common — rank-dependent ghost cells for any multi-rank wall-velocity BC run in pre/post. Fixed by adding the member broadcasts (simulation already had them).

2. muscl_eps was never broadcast, and its post-read derivation runs only under f_is_default(muscl_eps) with defaults assigned rank-0-only — so non-root ranks carried divergent values on every multi-rank MUSCL run. Fixed by broadcasting it like every other registered REAL.

3. 14 dead registrations crash the namelist read: lag_params%{T0, Thost, c0, rho0, x0} and fluid_pp%{mul0, ss, pv, gamma_v, M_v, mu_v, k_v, cp_v, D_v} were removed from the derived types (Refactor subgrid bubble models #1085/fix swapped vapor gas properties #1093) but never deregistered from the toolchain — setting any of them in a case file aborts with a misleading "datatype mismatch". Deregistered; the case validator now rejects them by name instead.

With the dead members gone, the fluid_pp/lag_params broadcast emitters convert from hardcoded member lists to registry walks — a member added to those types is now broadcast by construction, closing the gap that allowed bugs 1–2 to exist.

Verification

• Full golden suite (sharded, multi-rank) green on the final tip.
• Broadcast tuple-set comparison per target: deltas are exactly the intended additions (vb/ve members in pre/post, muscl_eps everywhere it is registered) and the dead-member removals; everything else identical.
• Toolchain pytest green, including new tests pinning the registry-walk emission and the deregistrations.

Issues fixed

Fixes #1559 (muscl_eps never broadcast)
Fixes #1560 (wall-velocity vb/ve members never broadcast in pre/post)
Fixes #1561 (pre_process broadcasts integer BC codes as mpi_p)
Fixes #1562 (14 phantom parameter registrations crash the namelist read)

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[sbryngelson]

Code review

Found 1 issue:

1. pre_process broadcast the integer bc_x/y/z%beg/%end members with mpi_p (an 8-byte real transfer over 4-byte integers — undefined behavior that happened to work by struct adjacency), while simulation and post_process correctly use MPI_INTEGER for the same members. Pre-existing on master, but this PR reorganized exactly this section — making it the fourth multi-rank broadcast defect this series fixes. Fixed in ea0f69b, which also adds a test pinning the hand-written vb/ve and BC-code residue so a merge conflict cannot silently drop them.

MFC/src/pre_process/m_mpi_proxy.fpp

Lines 44 to 47 in 740a29f

	& 'y_domain%end', 'z_domain%beg', 'z_domain%end', &
	& 'bc_x%beg', 'bc_x%end', 'bc_y%beg', 'bc_y%end', &
	& 'bc_z%beg', 'bc_z%end', 'bc_x%Twall_in', 'bc_x%Twall_out', &
	& 'bc_y%Twall_in', 'bc_y%Twall_out', 'bc_z%Twall_in', &

🤖 Generated with Claude Code

_{- If this code review was useful, please react with 👍. Otherwise, react with 👎.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 61.00%. Comparing base (ac30c32) to head (ca17a0d).
⚠️ Report is 7 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1553      +/-   ##
==========================================
- Coverage   61.17%   61.00%   -0.18%     
==========================================
  Files          74       75       +1     
  Lines       20313    19970     -343     
  Branches     2961     2930      -31     
==========================================
- Hits        12427    12182     -245     
+ Misses       5870     5806      -64     
+ Partials     2016     1982      -34     

☔ 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.