Postgres regression 7b

[LucaCappelletti94]

• Add Expr::XmlConcat(Vec<Expr>) variant with Display and Spanned implementations for PostgreSQL's XMLCONCAT(xml[, ...]) function
• Introduce maybe_parse_xml_function dispatcher in the parser, hooked at the top of parse_function, which routes XML function names to dedicated AST nodes on dialects where supports_xml_expressions() is true
• Guards on a single unquoted identifier with case-insensitive matching, so only bare xmlconcat(...) on PostgreSQL and Generic dialects takes the new path — all other dialects and qualified/quoted names fall through to the normal function-call parser

This is the second PR in a chain splitting #2252 (XML function support) into reviewable units. Builds on #2299 which introduced the supports_xml_expressions() dialect method.

[iffyio]

xmlconcat looks like a regular function syntax wise, is the parser unable to parse it today?

[LucaCappelletti94]

It can be parsed as a function, I added it to the enum nevertheless for easy of use, for the same reason that other functions like CEIL for instance appear there. Of course, open to discussion whether it is worth it.

[iffyio]

ah I see, I think we should keep the parsing as is that it gets represented as a regular function vs making it a special case

[LucaCappelletti94]

Ok so, just to be clear, I should close this particular PR and leave this as-is?

[iffyio]

Yeah we can close this and have the current behavior as is