Multibyte UTF-8 character split by newline in MANIFEST. MF

### Affected version

3.5.1

### Bug description

When writing manifestEntries containing multibyte UTF-8 characters these characters are split by newline producing 65533 character. This violates Jar File Specification and IETF RFC 3629. Specification demands UTF-8 character either on previous or next line. RFC defines encoding algorithm for character as choice of 1 to 4 octets, not allowing character to be treated as 1-octet only.
Same bug exists in OPENJDK class Manifest.java, may be it used by plugin causing bug in plugin.

Reproduce the issue with manifestEntries containing multibyte UTF-8 characters and inspect the plugin's manifest-writing path, comparing it with the mentioned OpenJDK Manifest.java behavior. Check the MANIFEST.MF line-wrapping rules in the Jar File Specification and RFC 3629. Done means UTF-8 characters remain intact across newline boundaries, with regression coverage for multibyte entries.

Written by the indexing model from the issue text.