fix(print): sequential clearance Kahn sort + above-rod Y extent (#630, #793)

[aceRage]

Description

Two-commit port of sequential print-by-object clearance fixes from Snapmaker/OrcaSlicer into Ultra. Only src/libslic3r/Print.cpp is changed.

This stops false "Assembly is too tall, and collisions will be caused" errors on MakerWorld/Bambu 3MFs whose object-list order is not a valid print order, and then tightens the Y-overlap test to the geometry that can actually hit the X rod.

Commit order (A then B)

1. Part A — fix(print): sequential clearance order via Kahn sort (#630)

   • Upstream: Snapmaker/OrcaSlicer#630 (open)
   • Replaces the disabled #if 0 score-propagation heuristic / #else object_index sort in Print::sequential_print_clearance_valid with a Kahn topological sort over clearance constraints.
   • At most one instance may need the last slot (height > height_to_lid). Y-overlap plus height > height_to_rod ⇒ must print before. Ties broken by object_index.
   • If the graph is unsolvable, fall back to object-list order so real collisions still error.
   • Vertical clearance check left unchanged in this commit.

2. Part B — fix(print): sequential clearance uses above-rod Y extent (#793)

   • Upstream: Snapmaker/OrcaSlicer#793 (open, dirty / stacked)
   • Logical change is only commit ebdb36dea99831778eb3cc04ffc61ad134683ab2 (~+63/−14 in Print.cpp).
   • Do not merge or cherry-pick the whole Sequential print: only material above height-to-rod can collide with the beam (stacked on #630) Snapmaker/OrcaSlicer#793 PR branch. GitHub currently shows ~+53k/−52k across 41 files because the branch is stacked/noisy (ASA/ABS, GCodeViewer, Flatpak, sidebar, flutter_web, etc.). Those files are not part of this port.
   • After Part A: add has_above_rod / above_rod_min_y / above_rod_max_y; compute above-rod Y from transformed mesh triangles (any vertex above rod contributes all three Ys; skip non-model_part volumes); use that extent in the Fix sequential print collision check depending on object list order Snapmaker/OrcaSlicer#630 constraint graph and in the vertical clearance loop. Tall-for-rod uses has_above_rod, not height > hc2.
   • Horizontal toolhead-radius check and height-to-lid last-slot rule are unchanged.

Screenshots/Recordings/Graphs

No UI change. Validation on Ultra hardware / MakerWorld 3MFs still needed locally (see Tests). Upstream Snapmaker#630/Snapmaker#793 include before/after slice screenshots and a printed above-rod test plate.

Tests

• Diff stats match the intended logical hunks: Part A +116/−94 (same as KuzuriAo 69f4a83 / Fix sequential print collision check depending on object list order Snapmaker/OrcaSlicer#630); Part B +63/−14 (same as ebdb36d).
• No existing Catch2 coverage for sequential_print_clearance_valid under tests/.
• Cloud compile/test limits: this environment cannot build Snapmaker_Orca / run ctest (wxWidgets deps, full slicer toolchain, and printer-profile validation are not available). Please compile locally and re-slice the MakerWorld/Bambu 3MFs from Fix sequential print collision check depending on object list order Snapmaker/OrcaSlicer#630 plus aboverod_test.3mf from Sequential print: only material above height-to-rod can collide with the beam (stacked on #630) Snapmaker/OrcaSlicer#793.
• Genuine unavoidable collisions should still produce the same "too tall" error (unsolvable graph falls back to object-list order).