Strange, that cargo_fmt did reformat the code that previously
was acceptable and didn't changed in the previous commit

… only if the deserializer has a bug

…r `visit_string` instead

It is responsibility of a `Visitor` to return an error if the provided data can not be handled

…l `deserialize_str` instead

Deserializer methods are only hints, if deserializer could not satisfy
request, it should return the data that it has. It is responsibility
of a Visitor to return an error if it does not understand the data

UnitDeserializer::new() available only since serde 1.0.139 (serde-rs/serde#2246)

Provide to visitor a string or a map instead.

Deserializer methods are only hints, if deserializer could not satisfy
request, it should return the data that it has. It is responsibility
of a Visitor to return an error if it does not understand the data

…e, tuple or struct from QName

Provide to visitor a unit instead.

Deserializer methods are only hints, if deserializer could not satisfy
request, it should return the data that it has. It is responsibility
of a Visitor to return an error if it does not understand the data

…:Syntax(SyntaxError)`

…sults of `read_to_end`

Co-authored-by: Daniel Alley <dalley@redhat.com>

…`DeError::UnexpectedEof`

…n read text in a tag and cover that case by test

…cases where Eof was received while waiting end tag

… cover case for a field of a sequential type

… explicitly show this using `unreachable!`

when we in that branch, `Start` or `Text` event already was picked

…no corresponding open tag

…te 2 different errors

… in cases where start tag is missing

…ed tags are not allowed even in HTML

…omment)`

The only usage was removed in 792d23d in tafia#445

Co-authored-by: Daniel Alley <dalley@redhat.com>