[DWAA1660]

What and why

Fixes #160. history::History::record truncated the redo steps and then called folds_into_last. That merges typing into the current step when it's the same Kind::Typing(block) within the 700 ms pause. Straight after an Undo, the current step is the state Undo had just restored. Typing then overwrote it in place, and the next Undo skipped past it. In the issue's sequence, typing abc, an edit, Undo, then X within the window, the final Undo returned "" instead of "abc".

record now notes whether there were redo steps (at + 1 < steps.len()) before truncating them. If there were, the user branched from an Undo and the new step never folds. It becomes a step of its own after the restored state. Contiguous typing with no Undo in between still coalesces as before.

How it was checked

• New history::tests::typing_after_an_undo_is_a_step_of_its_own:
  • the issue's sequence: after the new X, Undo returns abc, then "". Redo walks forward through abc → abcX on the new branch and stops.
  • a control: three keystrokes of ordinary typing are still a single step
  • Fails without the fix ("" instead of abc), and passes with it
• Existing history tests unchanged and passing
• cargo test -p notecraft-ui-egui: all pass (303 unit, 202 UI)
• cargo fmt --all -- --check, cargo clippy -p notecraft-ui-egui --all-targets -- -D warnings: clean

🤖 Generated with Claude Code