Measure representative publication projection costs

[10-X-eng]

Outcome

Records a reproducible 307-member VibeScript Assembly publication baseline and adds opt-in Tracy spans that isolate Tree and Feature Timeline item/icon costs. The trace proves the final 1.864-second heartbeat gap is the sequential Tree and Timeline refresh, not semantic projection or tessellation.

Risk

Low. This PR adds trace-only native scopes, a source contract test, and performance evidence. Tracy remains compile-time/opt-in as established by the parent PR; no public API, schema, preference, or default behavior changes.

Red/green and verification

• RED: C:\Users\robit\vibecad.pixi\envs\default\python.exe -m pytest src/Mod/VibeCAD/vibecad_tests/test_performance.py::test_native_projection_trace_separates_measured_hot_paths -q — failed with the three newly required Tree spans absent.
• GREEN: same focused command — 1 passed.
• C:\Users\robit\vibecad.pixi\envs\default\python.exe -m pytest src/Mod/VibeCAD/vibecad_tests/test_performance.py -q — 12 passed.
• Rattler CMake/MSVC incremental install target — passed and installed FreeCADGui.dll.
• �uild/run_vibescript_publication_307_native_trace.ps1 — correct 307-member publication; 322 document objects; native trace exported successfully.

Compatibility checklist

• Existing public functions/APIs remain present and compatible.
• No preference keys, tool names, or schema fields changed.
• Defaults preserve previous behavior.
• No deprecations.
• No breaking changes.

[10-X-eng]

Closed at the repository owner request; this performance implementation is being discarded.