fix: don't mark first launch complete on startup script error

[Copilot]

Any transient failure (network timeout, disk full, permission error) during first-launch asset retrieval permanently suppressed the startup script on all subsequent launches, leaving the user with missing assets and no recovery path short of manually deleting ~/.voxkit/.first_launch_complete.

Changes

• src/voxkit/gui/workers/startup.py
  • Removed mark_first_launch_complete() from on_error — first-launch flag is now only set on successful script completion, allowing automatic retry on next launch
  • Removed redundant from PyQt6.QtCore import QTimer re-import inside on_error (already in outer scope)
  • Replaced misleading comment # Still mark as complete to avoid running again with accurate intent

# Before
def on_error(error_msg: str):
    ...
    mark_first_launch_complete()  # silently bricks first-launch on any failure

# After
def on_error(error_msg: str):
    ...
    # Do NOT mark first launch complete on error — allow retry on next launch
    QTimer.singleShot(2000, loading_dialog.close_gracefully)

Testing

• No new tests added; the fix is in the control flow of a GUI worker (excluded from coverage by convention)

• Existing tests/storage/test_utils.py coverage of mark_first_launch_complete / is_first_launch remains unaffected

• Test testable logic

• Edge cases addressed

Testing Notes: The only behavioral change is that mark_first_launch_complete() is no longer called on error. Successful completion path is unchanged.

Documentation

• Code comments added/updated

Documentation Notes: Inline comment updated to reflect correct intent.

Checklist

• Code follows project conventions
• No breaking changes (or documented if intentional)
• All CI checks pass