parsing non-RTF content should throw exception

[stippi]

When parsing content that is not RTF, and when the content by chance contains no '{' characters, the low level parser will keep adding all chars to its character sequence cache. In the end, no hook function of the listener will be called and it looks the same to the upper layers as if an empty but otherwise valid RTF document was parsed. Any suggestions on how to best fix this? I can see multiple ways to approach this.