Add support for processing surrogate pairs in sequenced \uD800\uDC00 items

[harningt]

Surrogate pairs as described in the title will be decomposed into a pair of 3-byte utf-8 encoded chunks that are technically invalid. These should be decomposed into a 4-byte utf-8 encoded value for the represented codepoint.

[mtdowling]

+1, this would be awesome.