i18n-da: synchronized with
37a5a397f150
templater: cache full tree
store the parsed version of subtemplates in higher level parse lists, avoid
run-time cache lookups
templater: use recursive flattening
This avoids slow list operations compared to the iterative version