fix: Correct G92 E reset after deferred perimeters (issue #17)

[adele-with-a-b]

Absolute Extrusion Fix

In absolute extrusion mode (used by PrusaSlicer), the G92 E reset after deferred perimeters used myline.previous.e — the E value that includes the deferred perimeters' extrusion. But since those perimeters were replayed with relative extrusion (M83), the firmware's E register never advanced by that amount.

This caused a massive blob at the start of the next feature (outer wall) as reported in #17.

Fix

G92 E now uses self.last_noninternalperimeter_state.e — the E value from before the deferred perimeters. Applied to both generate_deffered_perimeters() and the kept perimeters path.

# Before:
buffer.append(from_gcode(f"G92 E{myline.previous.e} ..."))

# After:
e_reset = self.last_noninternalperimeter_state.e if self.last_noninternalperimeter_state else myline.previous.e
buffer.append(from_gcode(f"G92 E{e_reset} ..."))

Impact

• PrusaSlicer (absolute extrusion): Fixes blob at layer transitions
• Bambu Studio / OrcaSlicer (relative extrusion): No-op — these slicers use M83 by default

Closes #17