Thank you for the above John.
It seems like this works if it is a csv delimited import. I replicated your test and it was fine. But when I changed the delimiter to 'tab' I get the UDAs squashed together on one line.
it would be helpful if someone else could verify this behaviour as well.
I can change to csv delimiters if need be. But should I also raise an SR?
As it was a tab delimited file, I had to delimit my UDAs with a tab as well. eg.