This is a continuation of #205.

The developer ergonomics of building a gesture recognizer with pointer events now are painful, because there are too many things you need to know and remember to do it correctly. For instance:

• If you forget to set touch-action: none, pointermove won't be dispatched from a touch screen.
• If the user has a mouse/pen and starts a gesture on a link or image, dragstart will interrupt the pointer stream. To prevent this, the author must call downEvent.preventDefault(). This means the author needs to know and remember which behaviors are default for a down event and to be OK with preventing all of them.

Just these two already introduce a tremendous mental burden that make pointer events too hard to work with. Experienced developers may forget to do them, and newcomers probably don't even know about them. Without the luxury of both time and resources to be able to test a UI on all different types of pointers, an author might not even know there's a problem, but users will have a poor experience.

How can we make it easier for an author to opt-in to pointer events (and out of other behaviors like drag-and-drop)?