Add a SumEncoding option that's somewhere between a TaggedObject and an UntaggedValue

Sometimes the tag and the content of the JSON you're processing are on the same level where both `TaggedObject` and `UntaggedValue` will not suffice. I propose adding another `SumEncoding` option that handles this case. The implementation should be very similar to that of `UntaggedValue`.

Start by reviewing the existing SumEncoding options, especially TaggedObject and UntaggedValue, to understand their behavior and implementation. Define the same-level tag-and-content JSON shape with examples, then add the new option and tests showing that it handles that shape without breaking the existing encodings.

Written by the indexing model from the issue text.