fix(cli): surgical port of assemble-list crash + CLI extruder expand (#560/#562)

[aceRage]

Description

Surgical HIGH-CONFLICT manual port of two open Snapmaker Orca PRs onto Ultra. Manual hunks only — Ultra PartPlate.cpp / Snapmaker_Orca.cpp are not wholesale overwritten.

Upstream

• Snapmaker/OrcaSlicer#560 (OPEN) — CLI assemble-list crash during plate loading (null plater / GUI-only texture + raycaster paths)
• Snapmaker/OrcaSlicer#562 (OPEN) — CLI crash from GUI filament state during extruder expansion

Why this is HIGH conflict

Ultra CLI + PartPlate + Assemble Separately have drifted from Snapmaker Orca. This port adapts exact Snapmaker#560/Snapmaker#562 hunks onto the drifted Ultra sources and keeps Ultra CLI debug logs.

No AMP / profile / Stream refactors.

Changes

Part A — Snapmaker#560 (3 commits)

1. fix(cli): trim .stl name by 4 chars in assemble-list (#560)

   • STL-only: object_name.erase(..., -4) so names do not keep a trailing .
   • .obj remains -3
   • construct_assemble_list entry name unchanged

2. fix(partplate): null-safe plate name texture + invalidate (#560)

   • generate_plate_name_texture() early-returns when canvas/plater is missing
   • New invalidate_plate_name_texture() (declared in PartPlate.hpp)
   • clear() / set_plate_name call invalidate instead of raw m_name_texture.reset() / generate
   • Ultra adaptation: invalidate still resets m_name_texture so GUI lazy-regenerate via render_plate_name_texture keeps working (literal Fix CLI assemble-list crash during plate loading Snapmaker/OrcaSlicer#560 invalidate would leave get_id() != 0 and stall the displayed name)

3. fix(partplate): skip GUI set_shape work when m_plater == nullptr (#560)

   • Wrap logo / mesh / icon / grid / raycaster / name-texture work in if (m_plater != nullptr)
   • Keep non-GUI Ultra set_shape work outside the wrap (calc_bounding_boxes, height limit)

Part B — Snapmaker#562 (1 commit)

4. fix(cli): skip expand_plate_extruders in get_extruders_under_cli (#562)
   • Remove/skip expand_plate_extruders() only after unique-sort in get_extruders_under_cli
   • Comment left: needs GUI preset_bundle
   • GUI get_extruders helpers / dual-nozzle expand paths unchanged
   • Ultra CLI debug logs around the function retained

Files

• src/Snapmaker_Orca.cpp — one-line STL trim in construct_assemble_list
• src/slic3r/GUI/PartPlate.cpp + .hpp

Screenshots/Recordings/Graphs

N/A — CLI crash / null-safety; no UI layout change intended. GUI plate-name invalidate still lazy-regenerates on next render.

Tests

Cloud CLI smoke limits: this environment cannot run Snapmaker Orca CLI assemble-list or plate-load smoke (no built slicer binary, no printer/assemble fixtures). Verification here is surgical-diff review + static consistency checks:

• STL trim is .stl only; .obj still -3; function name unchanged
• expand_plate_extruders still called from GUI helpers (get_extruders, get_extruders_without_support); skipped only in get_extruders_under_cli
• set_shape still runs calc_bounding_boxes / calc_height_limit without a plater
• Ultra CLI BOOST_LOG_TRIVIAL debug lines around get_extruders_under_cli retained
• Code review: surgical port acceptable as-is; Ultra m_name_texture.reset() inside invalidate is required to avoid a GUI name-stale regression

Pre-existing (out of scope): estimate_wipe_tower_size() can still fall back to GUI get_extruders() in CLI if plate_extruder_size == 0. Not part of Snapmaker#560/Snapmaker#562.

Manual follow-up on a full desktop build (same as upstream Snapmaker#560/Snapmaker#562 validation):

• --load-assemble-list single-object control: no crash, G-code export
• Two-object same-plate assemble-list: no crash, G-code export
• GUI plate rename / clear still refreshes the plate name texture