Standardize initTouchEvent?

[RByers]

The WebEvents working group previously decided not to normatively specify initTouchEvent because there were differences in implementation and weird extra co-ordinate parameters that made no sense.

However, according to an httparchive search there is a non-trivial amount of usage of this API (1236 of the top 300k websites as of Aug 2014). For example: https://www.fieldandrurallife.com/ and http://www.arccjournals.com/. In both cases it looks like it's just being used as a form of touch feature detection.

However, Chrome UMA data indicates that initTouchEvent is virtually never called in practice on Chrome. I wonder why? We're tracking understanding/fixing initTouchEvent behavior in Chrome here

[RByers]

/cc @choniong

[RByers]

After doing more research, it seems that MOST of the uses of initTouchEvent we see in httparchive actually wouldn't work on Safari or Chrome (because Safari requires all 18 arguments to be supplied (despite their documentation saying otherwise) - perhaps a breaking change at some point that nobody noticed). We're now considering removing our (incompatible) initTouchEvent method from blink entirely. If we decide to do that, then I propose we undo the change to add it to the spec.

[RByers]

We've decided not to do this after all in favor of just pushing the constructors. Let me know if anyone disagrees - happy to discuss further.