bug fix - during call site transfer func we temporarily double nodes per allocation...

bug fix - during call site transfer func we temporarily double nodes per allocation site where nodes from the callee are called shadow nodes, and nodes that were in the caller before the call are named normal nodes.  During the transfer we move the shadow newest, which are the newest, into the caller and merge normal nodes down to make room.  The bug is that during the merge of normal nodes in this procedure there shoudl be a corresponding aging of reachability states, so when global sweep occurred after the node merging process some states look imprecise and get swept away