Coming back to this with fresh eyes I found the solution, my for loop has been changed to parse the cell array as follows
if true
% code
for i = 1:length(reqrefs)
reqids = reqs{i};
[nrows,ncols]=size(reqids);
for col=1:ncols
fprintf(fileID,'\t<XREF>\n\t\t<ID>%s</ID>\n\t\t<SYS>%s</SYS>\n\t</XREF>\n',char(reqids{:,col}),reqrefs{i});
end
end
The text is now correctly extracted from the model on multi-lines of the annotation.