[Flare] Ensure mouse events can use target to validate press

[trueadm]

This PR fixes an interesting bug that we encountered internally using the Press responder. When using mouse events (via pointer events or mouse events), we have a guarantee that the event target will be that of the one we're moving/releasing from (unlike touch events, which can have the same target as when the press was initiated). This mechanic for mouse events lets us bypass the getBoundingClientRect logic, which is:

• expensive as it requires a layout reflow
• can be incorrect in cases where children of the responderTarget are absolutely positioned
• can sometimes report width and/or height as 0 when some CSS styles are applied

This behaviour already existed in the current Press implementation. But there were two bugs:

• isTargetWithinNode wasn't taking into account the alternate fiber
• we weren't setting isPressWithinResponderRegion to true if we knew the mouse event target was in the responder region (by using the targets).

This fixes both cases, which also fixes the internal bug we were seeing.

[sizebot]

No significant bundle size changes to report.

Generated by 🚫 dangerJS

[necolas]

Why is this checking for only mouse pointers?

[trueadm]

That's what it was doing before. Touch event pointers have the same target until released, so if you remove the pointer type check, pointer events never deactivate, even if you move outside the target.