substr wrongly applied to utf8 strings

Using substr() on a UTF8 string that contains multibyte characters may produce an invalid UTF8 string.

For example, this code may produce an invalid string (will result in "warning: htmlspecialchars(): Invalid multibyte sequence in argument in /.../bootstrap.inc on line 857."):

    if (strlen($block_description) > 16) {
      $block_description = substr($block_description, 0, 16) . '...';
    }

and can be changed to this to solve the problem:

    if (drupal_strlen($block_description) > 16) {
      $block_description = truncate_utf8($block_description, 16, FALSE, TRUE);
    }

The bug can be reproduced with the following translation:

msgid "None"
msgstr "טרם ניצפה"

The attached patch makes this change for the three instances I could find where this potential problem is apparent.
There are other uses of substr() which are either not operating on a translated strings or which I was not sure about.