Check why UTF-8 decoding does not fail on illegal encodings of Surrogate codes

[cowtowncoder]

See FasterXML/jackson-dataformats-binary#65 for background: it looks like UTF-8 json parser is not checking for case of 3-byte character being surrogate: I think it should.
(although there is still the whole issue of why JSON escapes may be used in such a way...).