handle non-ASCII encodings of file names in tar archives

Summary:
BUG: 266141

As UTF-8 may use multiple bytes per character, it's required to measure encoded file name length in bytes, not just in characters. It's possible to accidentally truncate file name if its name in characters is shorter that 100, but in bytes — longer than 100. Also, checksum calculation must be performed on unsigned bytes. Otherwise bytes in range 0x80-0xff when promoted to int become negative.

Reviewers: dfaure, kossebau

Reviewed By: dfaure

Subscribers: xyquadrat, broulik, cfeck, ibragimovrinat, kde-frameworks-devel

Tags: #frameworks

Differential Revision: https://phabricator.kde.org/D14674