Based on my question here, I think it would be nice to have a transformer that could extract the character encoding of text attributes (e.g. "EncodingExtractor") into a new attribute. Besides (or instead of) a transformer, I guess it could also be useful to have this available as a feature function (e.g. @ExtractEncoding).
In order for this encoding attribute to be useful within Workbench, the AttributeEncoder transformer "Destination Encoding" field should then become attribute-driven. There are also implications for the Python API.