Fix handling of extraneous data in Unmarshal() & Valid()

[fxamacker]

Previously, Unmarshal() and Valid() ignored trailing bytes (if any) after the current CBOR data item. This matched the behavior of Decoder.Decode().

However, Unmarshal() and Valid() are typically used for single CBOR data item, so trailing bytes (if present) should return ExtraneousDataError given what RFC 8949 says:

A well-formed data item uses the initial bytes and the byte strings and/or data items that are implied by their values as defined in CBOR and does not include following extraneous data. CBOR decoders by definition only return contents from well-formed data items.

This is a bug fix rather than a new non-default option because the previous behavior of ignoring extra data didn't exactly match RFC 8949 and was also different from how Marshal() in encoding/json handles extraneous data.

Special thanks to @zensh for reporting this issue and proposing a solution.

Closes #359

How to handle extraneous data as non-errors

Decoder.Decode() and UnmarshalFirst() can handle extraneous data as non-errors.

• Decoder.Decode() uses io.Reader and ignores remaining bytes after reading a single CBOR data item. By design, it does not treat extraneous data as an error.

• 🆕 UnmarshalFirst() will decode first CBOR data item and return trailing bytes (if any). By design, it will not treat extraneous data as an error.

UnmarshalFirst(data []byte, v interface{}) (rest []byte, err error)

Decoder.Decode() and UnmarshalFirst() are also useful for decoding CBOR Sequences (RFC 8742).