Implement mergeplanar: merge coplanar triangles into planar faces

[slugdev]

The flag was parsed but never used (and printed a wrong message while consuming the next argument). It now merges edge-connected coplanar triangles into single planar faces:

• Union-find over edge-adjacent triangles with parallel normals; requiring adjacency keeps disjoint coplanar patches as separate faces (global plane binning is what broke the earlier merge_planes attempt)
• Directed boundary edges chained into closed loops, preserving STL winding so hole loops stay correctly oriented within one face
• Vertices and boundary edge curves shared globally so neighboring faces reference identical topology; open chains are dropped with a warning instead of emitting invalid loops

Also fixes the arg parser consuming the argument following mergeplanar and documents the option in help/README.

Tests: cube content check (12 tris -> exactly 6 faces/12 edges/8 verts, unmerged path unchanged) plus bucket and cat_dish smoke runs; bucket merges 11286 -> 6236 faces with no dropped chains. The *.cmake gitignore catch-all was removed so the test script can be tracked.

[Copilot]

Pull request overview

Adds a new mergeplanar CLI option to merge edge-connected coplanar STL triangles into single planar STEP faces, improving output topology for planar regions and fixing previously broken argument parsing for the flag.

Changes:

• Implement planar triangle merging in StepKernel via adjacency-based union-find and boundary loop extraction.
• Wire up the mergeplanar CLI flag, update usage/help text, and add regression/smoke tests.
• Track the new CMake test script by removing the *.cmake ignore rule.

Reviewed changes

Copilot reviewed 7 out of 8 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
test/cube.stl	Adds a unit-cube STL fixture (12 triangles) for mergeplanar regression testing.
test/check_mergeplanar.cmake	Adds a CMake script to assert expected entity counts with/without mergeplanar.
StepKernel.h	Extends build_tri_body signature with merge_planar flag and declares merged implementation.
StepKernel.cpp	Implements build_tri_body_merged and routes build_tri_body based on merge_planar.
README.md	Documents the mergeplanar option in usage (needs alignment with other existing CLI options).
main.cpp	Fixes mergeplanar parsing (no longer consumes next arg) and passes flag to StepKernel.
CMakeLists.txt	Adds mergeplanar_* tests (cube regression + bucket/cat_dish smoke runs).
.gitignore	Removes *.cmake ignore so the new test script can be committed.

---

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

[slugdev]

@copilot resolve the merge conflicts in this pull request

[Copilot]

@copilot resolve the merge conflicts in this pull request

Resolved the merge conflicts. The three conflicting files were handled as follows:

• CMakeLists.txt: Kept all tests from both branches — the three new mergeplanar_* tests and the edge_curve_line_geometry test from master.
• README.md: Used master's polished formatting while adding the mergeplanar option description.
• main.cpp: Kept the fix from this branch (no spurious arg_cnt++ in the mergeplanar case) and incorporated the refactored units/schema handlers from master.

All 11 tests pass after the merge. Commit: Resolve merge conflicts with origin/master

[Copilot]

Pull request overview

Copilot reviewed 7 out of 7 changed files in this pull request and generated no new comments.

Suppressed comments (2)

StepKernel.cpp:228

• The directed-edge map overwrites earlier owners when the same directed edge appears multiple times (non-manifold input). That can silently drop triangle adjacency and lead to incorrect grouping/boundary extraction; it’s better to detect duplicates and warn (or handle) instead of overwriting.

	for (std::size_t t = 0; t < tri_cnt; t++)
		for (int c = 0; c < 3; c++)
			directed_edge_tri[make_edge_key(vert_of(tri_ids[t], c), vert_of(tri_ids[t], (c + 1) % 3))] = t;

StepKernel.cpp:367

• These newly added lines appear to include stray carriage-return characters (mixed line endings), which can cause noisy diffs and tooling/lint issues. Please normalize this block to use consistent LF line endings.

		// Orthonormalize ref against the normal (matches the non-merged path's csys construction)

		double dotnr = ref[0] * n[0] + ref[1] * n[1] + ref[2] * n[2];

		ref[0] -= dotnr * n[0];

		ref[1] -= dotnr * n[1];

		ref[2] -= dotnr * n[2];