Fix Unicode support

[sammdot]

– All characters represented with Unicode escapes should also use placeholder characters for non-supporting RTF readers: e.g. my Luminex II translates "San José" San Jos\u233 as "San Jos", ignoring the e entirely
– Unicode characters from U+8000 to U+FFFF should use negative numbers
– Unicode characters above U+FFFF should be represented in UTF-16 encoding

Reference: https://www.zopatista.com/python/2012/06/06/rtf-and-unicode/