Morse code error handling

[ratijas]

<ITU as Dialect>::encode_unknown should really handle encoding errors instead of being an empty slice of code points.

This is our current implementation:

    fn encode_unknown(&self) -> KnownCodePoints {
        Self::EMPTY
    }

According to the Wikipedia, the protocol defines the following sequence for error under the Prosigns section:

· · · · · · · · 

Not sure if it is applicable for encoding unknown characters though.