ignore content-type encoding when choosing a parser for topic data

Now it will properly strip any encoding from content-type header before
comparing against available parsers.