dillon1000/react

Rewrite SyntheticEvent tests using public APIs only (#11525)

* generate synthetics events using public API

* rewritten createEvent to use public APIs
* removed all references SyntheticEvent.release

In order to test under realistic circumstances I had to move
the expectations into a callback in mosts tests to overcome
the effects of event pooling.

* run prettier

* remove empty line

* don't use ReactTestUtils

* run prettier and fix linter issues

* remove duplicate test

* remove invalid calls to expect

The removed `expect` calls verified the correct behaviour based on
missing `preventDefault` and `stopPropagation` methods.
The was correct as we used plain objects to simulate events.
Since we switched to the public API we're using native events which
do have these methods.

* set event.defaultPrevented to undefined

This was missed when the test was first migrated.
When emulating IE8 not only has returnValue to be false.
In addition defaultPrevented must not be defined.

* run all tests and format code

* rename instance variable to node

* remove backtick

* only simulate IE in normalisation test

* include assignment in definition

* add missing `persist` test

* use method instead of field to prevent default

* expect properties to be unchanged on persisted event

* optimise tests that deal with event persitence

* declare and assign `event` on the same line if not reassigned later