Return the correct charset in content-type

It is found that if the file is not UTF-8 encoded, then the default UTF-8 encoding of koa.js will be used.

Please return the correct charset of the file here

No file or test is named. Start by locating the static-cache middleware's content-type handling and reproduce the issue with a non-UTF-8 file. Done means the response reports that file's correct charset instead of always defaulting to UTF-8.

Written by the indexing model from the issue text.