Skip second _find_key for ReadWrite

[bpmckinnon]

I'm not sure about the internals for how memory is flagged as available, but right now when using the ReadWrite lock if the switch_to_unique returns true, then _find_key is run again. Would it make sense to
store the capacity (or pointer) of the set
do _find_key
if offset is valid
switch the lock to writer
if the capacity (or pointer) has changed or the entry at offset no longer matches
repeat _find_key
Then you wouldn't have to repeat the find unless your data has been invalidated.

[greg7mdp]

That's an interesting idea, but I'm not sure it would be much faster. I think the overhead for doing the find again is quite low.