Zinc adaptor in Pharo always defaults to UTF8 decoding of requests even if other encoding is set

[jbrichau]

• Set the Zinc adaptor's encoding to something other than UTF8 (e.g. Latin1)
• Go to http://localhost:8080/tests/functional/WAEncodingFunctionalTest
• Enter é in the 'GET' input field and push the submit button
  => decoding error

This is because the Zinc adaptor will always rely on Zinc for decoding of the bytes in the request. It defaults to UTF8 (in ZnPercentEncoder) and does not use the codec setting of the adaptor. In the scenario above, the page is served as latin1 and the browser will percent-encode the form values using latin1 codec in the request. The e is therefore encoded as %E9. On the server, Zinc tries to do: ZnPercentEncoder new decode: '%E9' which errors because it tries to use the UTF8 character decoder. (UTF8-based percent-encoding of é is %C3%A9)

I guess everyone is now using UTF8 anyways, so question is: do we need to fix this issue or do we need to remove the ability to work with anything else than UTF8?