The problem is with how email sending application split the file name parameter of Content-Disposition header:
Content-disposition: attachment;
filename*0*=utf-8''%CE%95%CE%B8%CE%BD%CE%B9%CE%BA%CE;
filename*1*=%B7%CC%81%20%CF%80%CE%B9%CF%83%CF%84%CF%89;
filename*2*=%CF%84%CE%B9%CE%BA%CE%B1%CC%81%2015-10-2; filename*3=012.pdf
National characters in UTF-8 encoding, take more then one byte. The first split is not done on a character boundary, but in the middle of a 2-byte character:
..................................................%CE;
filename*1*=%B7......................................;
CEB7 is 2 byte hex representation of a single character ('θ'). I'm not 100% sure if this is allowed by RFC2231.
Similar situation occurring with 2 adjacent encoded words is for sure not allowed, but it is recognized by Mail.dll.
We'll improve the parser to handle such situations.
[Edit]
The latest version handles such emails correctly.