Do not overwrite specified encoding of signed parts
Consider the (stripped down) message added as a new test-case.
Previously, after parsing the signed part, the charset would be overwritten to be UTF-8, resulting in broken display of the umlauts in the signature.
The updated version of this patch ensures that the charset related check in testOpenPGPInline continues to pass, too. I am not quite sure that this behavior actually conforms to RFC4880, which specifies that ASCII armored PGP blocks may specify their own encoding, and should be assumed as UTF-8, otherwise (section 6.2). However, I assume it is a rare corner case, and at least some version of KMail must have expected the given behvior.
Edited by Thomas Friedrichsmeier