When adding many objects(10.000+) via python at once, the notifier queue
becomes very large. Measured with hotspot, this duplication check takes a
huge amount of time.

This (draft) prototype adds a hashmap for looking up duplicates. Using this
hashmap i.o. the listbase for duplication check speedups the performance greatly.

This patch is a draft, meant to discuss the solution of using a different, more
robust datatype, for duplication check. I didn't take implementing the notifier queue
as a different, more robust data type into account as this might make the impact greater.
But I might imagine this can be a possibility as well i.o. introducing another variable for reading
(2 variables to keep in sync ).
Also this patch misses implementation of deleting/removing notifications and adding events.