This came up during refactoring of core.js into smaller pieces (including context_menu.js).

Context.contextFromPoint uses document.elementFromPoint(x, y). The elementFromPoint_() method in __gCrWeb implements similar functionality, and can also look into frames and iframes. However, the comment also suggests that the latter was added for iOS4/5, so it's not clear if this still adds value over the latest native implementation.

We should:
- Understand the difference between the latest native implementation and the __gCrWeb version.
- Evaluate if it makes sense to unify the two, and if so:
- What is the best path