QDomText mis-encodes characters that require surrogate pairs

The XML spec says¹ that

Characters referred to using character references MUST match the production for Char.

And the Char production² reads

Char	   ::=   	#x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]	/* any Unicode character, excluding the surrogate blocks, FFFE, and FFFF. */

eddy found, though, that qdom.cpp:encodeText() encodes surrogate pairs as two character references, each expanding to a surrogate pair code point; exactly what's expressly forbidden by the spec.

¹ https://www.w3.org/TR/REC-xml/#wf-Legalchar
² https://www.w3.org/TR/REC-xml/#NT-Char