refactor(reader): misc simplifications

[jvoisin]

• There is no need for getEncoding to return a string instead of an array of bytes, so let's make it return a []byte instead of a string.
• There is also no need to assign a decoder.CharsetReader when the encoding is utf-8, so let's not do that anymore.
• Moreover, there is no reason why the function used for decoder.CharsetReader has to be defined as a lambda instead of a proper function. One might argue the other way around, but a lambda is living on the heap, while a "real" function doesn't.

[fguillot]

There is also no need to assign a decoder.CharsetReader when the encoding is utf-8, so let's not do that anymore.

That one might need more testing. I don't recall all the details, but it's possible that it was done this way to handle some broken feeds (mismatch between the file encoding and the encoding declared in the XML file).

[jvoisin]

Then let's not do it for now :D