PieceTable: binary-search piece lookup; lazy line-start shift

Each edit shifted every later line start in the line index, O(lines) per
keystroke (0.5 ms near the top of a 1M-line file). The shift is now kept
pending ("from entry k, add d") and applied on read; further edits at the
same line only adjust it, and it is folded in when an edit lands
elsewhere.

locate() walked the piece list for every lookup (two per rendered line
per frame on an edited buffer); it now binary-searches prefix sums
rebuilt once per modification (keyed on version_, which every mutation
bumps; AdoptOriginal now bumps it after installing its piece).

Typing on line 10 of a 1M-line file with a frame's worth of line reads
per keystroke: 0.52 ms -> 0.017 ms. A new randomized test checks every
read path against a string model with consolidation forced; it fails
under each of three deliberate breakages of the new bookkeeping.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015rcVQiiWyTzoNrsvLSaAfB

PieceTable.h

@@ -237,9 +237,34 @@ private:
 	mutable std::uint64_t version_ = 0;
 	std::size_t total_size_        = 0;
 
+	// End offset of each piece (prefix sums) for binary-search lookup,
+	// valid while piece_ends_version_ == version_ (every mutation bumps it).
+	mutable std::vector<std::size_t> piece_ends_;
+	mutable std::uint64_t piece_ends_version_ = std::numeric_limits<std::uint64_t>::max();
+
 	// Cached line index: starting byte offset of each line (always contains at least 1 entry: 0)
 	mutable std::vector<std::size_t> line_index_;
 	mutable bool line_index_dirty_ = true;
+	// A shift not yet applied to line_index_: entries from index
+	// line_shift_from_ on are really line_index_[i] + line_shift_by_ (mod
+	// 2^64, so it may be negative). An edit moves every later line start;
+	// applying that at once cost O(lines) per keystroke, so consecutive edits
+	// on one line only adjust the pending shift.
+	mutable std::size_t line_shift_from_ = std::numeric_limits<std::size_t>::max();
+	mutable std::size_t line_shift_by_   = 0;
+
+	// Start of line i (i < line_index_.size()), with the pending shift.
+	[[nodiscard]] std::size_t lineStart(std::size_t i) const
+	{
+		return i >= line_shift_from_ ? line_index_[i] + line_shift_by_ : line_index_[i];
+	}
+
+
+	// Index of the first line starting after byte_offset.
+	[[nodiscard]] std::size_t lineUpperBound(std::size_t byte_offset) const;
+
+	// Fold the pending shift into line_index_.
+	void applyLineShift() const;
 
 	// Heuristic knobs
 	std::size_t piece_limit_             = 4096; // trigger consolidation when exceeded