fix: reparent deck onto string-prefix sibling instead of no-op

[krMaynard]

Linked issue (required)

Closes #5183

Summary / motivation (required)

reparented_name() used target.0.starts_with(&self.0) to detect a drop onto the dragged deck or one of its descendants. Because this compared the raw \x1f-joined names rather than component boundaries, a sibling whose name merely shares a string prefix (e.g. dragging foo::bar onto foo::barbaz) was misclassified as a descendant and the drop became a silent no-op.

The fix compares on \x1f component boundaries so only the deck itself or a genuine descendant is treated as a no-op.

Steps to reproduce (required, use N/A if not applicable)

1. Create decks foo::bar and foo::barbaz.
2. In the deck list, drag foo::bar onto foo::barbaz.
3. Observe nothing happens (silent no-op) instead of bar becoming foo::barbaz::bar.

How to test (required)

Checklist (minimum)

• I ran ./ninja check or an equivalent relevant check locally.
• I added or updated tests when the change is non-trivial or behavior changed.

Details

Adds a regression assertion to the drag_drop test. The full CI workflow (check on Linux/macOS/Windows, format, minilints) was run against this exact commit on my fork: https://github.com/krMaynard/anki-fork/actions/runs/29983583656

Before / after behavior (optional)

Before: reparenting onto a string-prefix sibling is a silent no-op. After: it reparents correctly.

Risk / compatibility / migration (optional)

Low risk.

UI evidence (required for visual changes; otherwise N/A)

N/A

Scope

• This PR is focused on one change (no unrelated edits).