Fixes #4886

In Crystal, handling multi-byte sequences in UTF-8 requires understanding that slicing by bytes can lead to invalid sequences if the slicing isn't aligned with character boundaries. In this case, attempting to slice a string by bytes can cut through multi-byte UTF-8 sequences, leading to invalid sequences.

To avoid this, strings should be sliced based on characters rather than bytes.

Explanation:

1. str.chars: Converts the string to an array of characters, which allows you to handle slicing based on characters rather than bytes.
2. chars[start_index, length]: Slices the array of characters, which is safe for UTF-8 as it ensures no multi-byte sequences are broken.
3. sliced_chars.join: Joins the sliced characters back into a string.

This approach ensures that the slicing respects the boundaries of UTF-8 characters and avoids invalid byte sequences.