I noticed this old thread when a best answer was awarded just now. There's some newer information that may be relevant.
The 2019 GEDCOM 5.5.5 specification (available as a PDF from here) says on page 61:
Children within a family group should be ordered chronologically, i.e. in order of birth, and not anti-chronologically, and the order of multiplets born on the same day should be preserved.
It qualifies that by saying user preferences should be respected. But since WikiTree already displays children in chronological order of birth, without user options, it seems worthwhile preserving chronological order among the CHIL subrecords of the FAM records. This seems at least mainly to be done at present. I don't think the standard requires that the INDI records be in chronological order within the GEDCOM file.
There are some other areas in which the files exported from WikiTree are not in compliance with the standard. The file encoding is not UTF-8 or UTF-16 with BOM. There's an illegal blank line, and illegal trailing spaces on value-less NOTE and CONT lines. Some lines are longer than the 255-character maximum. Profile comment text can appear in the file as raw lines, without any GEDCOM tag. Where the @ character appears, e.g. in email addresses, it is not doubled to @@ as required. There are "1 BIRT" records with no value and no subrecords. There are cases where HUSB or WIFE is in a FAM record, but the corresponding INDI record has no FAMS subrecord. The WWW subrecord is misused, for WikiTree link instead of after an ADDR subrecord; a NOTE or a REFN could legally be used for this URL instead.