fix: Restore layer Z after spiral-vase raft tool changes

[mvanhorn]

Summary

Update the spiral-vase path in GCode::travel_to to restore the writer to the requested/nominal layer Z when an earlier tool change or injected G-code left it at a different height, then retain the existing XY-only travel behavior once Z is correct. Base the decision on the generator's tracked position and the travel's target Z rather than on X2D-specific commands, raft counts, or a hard-coded clearance value, so other custom tool-change sequences receive the same correction. Preserve the no-extra-Z-move behavior for ordinary spiral layers where the tracked and nominal heights already agree, avoiding changes to seam blending and spiral interpolation.

Test plan

• Happy path: slice a hollow single-object spiral-vase print with two raft layers on filament 2 and the model on filament 1, with tool-change G-code ending at max_layer_z + 3; verify the first model extrusion after the raft is at the nominal first-object-layer Z.
• Edge case: slice the same spiral-vase/raft geometry without a Z-changing tool change; verify no redundant layer-height correction is introduced and the existing spiralized Z progression remains continuous.
• Error path: use custom tool-change G-code that moves to a clearance Z and leaves the current XY position unknown; verify the generated travel safely restores Z before any model extrusion and does not emit extrusion at the clearance height.

Why

On an X2D print that uses an auxiliary filament for a two-layer raft and the main filament for a spiral-vase object, the first object extrusion is emitted roughly ten layers above its intended position. The raft-to-object filament change runs custom machine G-code that may finish at a clearance Z such as max_layer_z + 3, and the generator records that external Z position. The spiral-vase branch of GCode::travel_to then emits only an XY travel, unlike the normal branch that travels in XYZ, so it does not restore the nominal object-layer Z before extrusion. The issue is limited to the interaction between raft-driven tool changes, externally moved Z, and spiral-vase travel behavior; ordinary spiral interpolation and the X2D profile template should remain unchanged.

Closes #11961

[Diatom-Bambu]

Here's a small suggestion: the action of descending the z-height should more reasonably be placed after the two lines (xy travel)
_"for (size_t i = 1; i < travel.size(); ++i)
gcode += m_writer.travel_to_xy(this->point_to_gcode(travel.points[i]), comment, use_short_travel_accel);"
（line 7728 below）
This is to ensure that the xy travel is performed at a higher height, and printing starts directly after descending, reducing the possibility of being scratched by the warped edges.

[Diatom-Bambu]

Hello, thank you for your contribution.
The problem indeed need a fix, but the action of descending the z-height should more reasonably be placed after the xy travel, which make it far from scratch.

[mvanhorn]

You were right about the ordering, and it is fixed in 5ebea7f.

The XY travel loop now runs before the target-Z calculation and the Z restoration, so the toolhead finishes its horizontal move and only then descends, instead of dropping to the layer Z first and travelling at that height.

It is a two-line reorder in travel_to; nothing else changed. clang-format reports no violations on the file. I could not run the full build here since it needs prebuilt dependency prefixes I do not have locally, so the CI build is the real check on it.

[Diatom-Bambu]

thanks for your fixing