lutra-text gets Buffer on a ropey rope: byte/char/UTF-16/line counts,
byte offset <-> (line, column) in UTF-16 code units or chars, absolute
UTF-16 offsets, atomic Edit batches in LSP TextEdit[] semantics, and
\n / \r\n / \r detection with a keep-the-file's-ending save policy.
All conversions return TextError instead of panicking.
ropey 1.6.1 (MIT) without unicode_lines, so only the LSP/CodeMirror
line breaks split lines. proptest moves into workspace deps.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>