org.apache.http.entity.ContentType#get method is too strict regarding malformed Content-Type headers

One HTTP server serving up ZIP/JAR file sends this header in it's response:

Content-Type: application/zip; charset=binary

While this is malformed header value (according to rfc2068 "Applications SHOULD limit their use of character sets to those defined by the IANA registry", and IANA does not define "binary"), HttpClient should be more robust IMO.

In this case, it might accept the mime type application/zip only and just neglect the charset (that results in java.nio.charset.UnsupportedCharsetException: binary anyway), since application/zip is not a text subtype.