See T141922#2578626 for context.
With hovercards enabled and the experiment enabled, users in the control group (previews OFF) when hovering a link multiple times and then clicking it will submit multiple click events.
=What=
In the hovercards OFF case, when the user clicks a link that has been hovered multiple times, then multiple click events are logged **with different `linkInteractionToken`s**. Events have a decreasing `totalInteractionTime`, e.g. in the automated tests that were run we saw one event with a `totalInteractionTime` on the order of 100ms and one on the order of 10ms (in real data this will depend on the user's actions and their timing).
=Correcting analysis=
These events can be de-duplicated by grouping click events by `pageTitleHover` and by `pageToken` but only for `action: 'opened in same tab'` (`opened in new tab` can send multiple click events legitimately, so no fix for that). The event with the lowest `totalInteractionTime` should be accepted (as it should be the last one) and all others rejected.
With such grouping we should see fewer click events on the OFF case.
=Why=
When preview is disabled a click event is bound in every hover that is never unbound, which doesn't happen when preview is enabled. As @phuedx mentioned in T141922#2574587, binding click events follows different code paths for the OFF and ON groups.
=In source=
* Click event bound in every hover when hovercards disabled: https://github.com/wikimedia/mediawiki-extensions-Popups/blob/master/resources/ext.popups.targets/desktopTarget.js#L158-L162
* Contrast with off+bind in every hover when hovercards enabled: https://github.com/wikimedia/mediawiki-extensions-Popups/blob/master/resources/ext.popups.renderer/desktopRenderer.js#L118
---
==AC==
* When hovercards is enabled but I'm in the control group (previews disabled), when hovering a link multiple times and then clicking it only one click event will be sent to analytics about that link I clicked.