decoding badly encoded strings

Would it be possible to correctly decode badly encoded strings like the following example ?

Header in received mail:
`Subject: =?UTF-8?B?KCBSZXF1ZXN0OkEsMzE4ODEsOTI5MTMgKSBJbmZvcm3D?=
=?UTF-8?B?oWNpYSBrIHBvxb5pYWRhdmtlIFVQVlMgxI0uIDkyOTEz?=`
It's badly encoded because it splits two byte character - first byte is at the end of the first encoded text and the second byte is at the beginning of the second encoded text.
I gets decoded to:
`( Request:A,31881,92913 ) Inform??cia k požiadavke UPVS č. 92913`
Instead of:
`( Request:A,31881,92913 ) Informácia k požiadavke UPVS č. 92913`

Reproduce the supplied Subject header and compare the decoded result with the expected Slovak text. Start by locating the Jakarta Mail MIME encoded-word decoding entry point; investigate how adjacent encoded words handle a UTF-8 character split across boundaries. Done means the example decodes without replacement characters and existing decoding behavior remains intact.

Written by the indexing model from the issue text.