Fix CLI assemble-list crash during plate loading

[Ohmicaudio]

Summary

This fixes a Snapmaker Orca CLI crash when using --load-assemble-list.

The crash reproduced locally in Snapmaker Orca V2.3.4 even with a single-object assemble-list control. Upstream OrcaSlicer V2.4.1 can slice the same same-plate assemble-list workflow successfully, so this appears to be a Snapmaker fork CLI assemble-list issue rather than an AMP/profile issue.

Changes

• Hardened CLI assemble-list plate loading so it does not initialize GUI-only plate/render/raycaster/name-texture state when there is no GUI plater.
• Fixed STL object name trimming in the assemble-list path.
• No slicing behavior changes are intended outside the CLI assemble-list loading path.

Changed files

• src/Snapmaker_Orca.cpp
• src/slic3r/GUI/PartPlate.cpp
• src/slic3r/GUI/PartPlate.hpp

Validation

Built successfully locally.

CLI tests after the fix:

• Single-object assemble-list control: exit code 0, G-code exported.
• Two-object same-plate assemble-list: exit code 0, G-code exported.

Generated G-code paths from local validation:

• B:\ohmic\Snapmaker-OrcaSlicer-cli-fix\outputs\amp_cli_patch_probe\single_control_cli_safe_final\plate_1.gcode
• B:\ohmic\Snapmaker-OrcaSlicer-cli-fix\outputs\amp_cli_patch_probe\same_plate_cli_safe_final\plate_1.gcode

Scope

This branch contains only the CLI assemble-list fix.

It does not include AMP planner code, benchmark docs, generated outputs, submission docs, profile changes, or generated files.

No production profiles were changed.
No G-code generation logic was changed.
No Snapmaker validation or safety logic was changed.

Motivation

This makes CLI-driven slicing and benchmark automation possible for workflows that currently require manual GUI slicing. It also avoids a hard crash when CLI assemble-list loading is used without the GUI plater state available.

[adenta]

Independent integration check on current upstream f34a62d: I built an arm64 Release CLI with the exact #560 commit (6fbafd5), plus the complementary existing fixes from #562 and #761 and the vendor resolver in #803. A Snapmaker U1 / 0.20 Standard / 20 mm cube CLI slice completed with exit 0 in 2.07 s; the representative BBL regression slice also completed with exit 0 in 2.07 s.

This is evidence for the combined headless path and compatibility with #803, not a claim that the timing or complete result is attributable to #560 alone.