events: optimize various functions

Cache events and listeners objects where possible and loop over
Object.keys() instead of using for..in. These changes alone give
~60-65% improvement in the ee-add-remove benchmark.

The changes to EventEmitter.listenerCount() gives ~14%
improvement and changes to emitter.listeners() gives
significant improvements for <50 listeners
(~195% improvement for 10 listeners).

The changes to emitter.emit() gives 3x speedup for the fast
cases with multiple handlers and a minor speedup for the slow
case with multiple handlers.

The swapping out of the util.is* type checking functions with inline
checks gives another ~5-10% improvement.