Clients should be disabling history and forms records together as they are behind the same check box in the UI. However, in a recent server side sample of disabled collections the two are apparently being disabled independently. See: https://bugzilla.mozilla.org/show_bug.cgi?id=1319923#c3
I wrote a little gist to try and see what our telemetry says about this. It looks at 1940631 pings - many of which has a large number of individual syncs (it doesn't show the total number of syncs, just pings). Of these, 12 syncs record a forms sync without history or vice-versa. The gist shows these sync records. https://gist.github.com/mhammond/2435863d482079931685d892dd42b3fc In hindsight I should have considered the UID, as we seem likely to find that they are all from a small number of users. However, the numbers seem low enough that it seems difficult to justify spending too much time digging in too far. Although often little things like this end up demonstrating something far more interesting. If necessary, we can dig more after hawaii.
(In reply to Mark Hammond [:markh] from comment #1) > In hindsight I should have considered the UID, as we seem likely to find > that they are all from a small number of users. heh - we do have uids (ironically, Thom is about to land a patch that means we wouldn't have for newer pings...) But as you can see in that gist, most uids are our "unknown" value of "0".repeat(32), and a few apparently valid other values. I wish I had a theory for what all this might mean...
Based on comment 11, we don't think this is worth investigating further.