Thanks for you quick reply Billiam :)
By saving and viewing the generated html I can confirm that even non-unique headings get unique generated IDs. However I did notice that the 2nd heading is assigned 1, the 3rd heading is assigned 2, and so on. So we're always off by one which is a bit awkward. It would be more natural to use (for me, at least) if the 2nd heading was assigned 2, the 3rd heading assigned 3, and so on.
Meanwhile, this works within a file, but I'm less confident that generated IDs would work for duplicated headings across files? Besides, having to update the target of existing links if ever a non-unique heading is added or removed from the project (or if ever a heading is edited?) seems less than ideal. Happy to learn otherwise, but perhaps this is where custom IDs would be a useful addition?
FWIW, I also noticed that the syntax checker complains about non-unique heading text. Is there an easy way to turn this rule off?