API Reference/modules/LrXml.html#xmlDomInstance:transform discusses normalizing the result you get back, and provides a snippet of string manipulation to do so:
"... apply a substitution pattern such as this: string.gsub( result, "[\r\n ]+", " " ) ..."
string.gsub( result, "[\r\n ]+", " " )
I /think/ this substitution is supposed to be "[\r\n\t ]+"
It looks like there is a literal tab character in the rendered content of the documentation, which somehow ate the '\t' escape sequence.
Retrieving data ...