Fix lexer::scan() and utils::impl MyStrExt for &str

[hardglitch]

Fix UTF-8 safety in string slicing

Problem:
The original code indexed &str using byte offsets (bytes[i] as char and slices like &s[a..b]), which breaks UTF-8 when multi-byte characters are present (e.g., Cyrillic). This caused panics like: byte index N is not a char boundary

Solution:

• Loops rewritten to use char_indices() so all slice indices always land on valid UTF-8 character boundaries.
• Slices computed safely with i and c.len_utf8().
• Functions scan() and split_first_whitespace() now handle Unicode correctly.
• Added is_only_whitespace() implemented via chars().all(|c| c.is_whitespace()).

Effect:
The library no longer panics on RTF containing Cyrillic or other multi-byte characters, while preserving correct behavior for ASCII.

[d0rianb]

Thank for the work,
Just a few comments on linting belwow

[d0rianb]

This could fix #18.
However, a test on the parser is not passing. It seems that a whitespace is missing.
maybe you could try this test in the lexer :

#[test]
fn should_lex_unicode() {
  let rtf = r#"{\u21834  \u21834 }"#;
  let tokens = Lexer::scan(rtf).unwrap();
  assert_eq!(
      tokens,
      vec![OpeningBracket, ControlSymbol((Unicode, Value(21834))), PlainText(" "), ControlSymbol((Unicode, Value(21834))), ClosingBracket]
  );
}