Container id problem

There is a problem with the way that we currently generate container ids. Like this:

1. process 1 reads user, it gets assignemnt(id=1,target=A)
2. process 2 modifies user: deletes assignemnt(id=1,target=A), adds assignemnt(target=B). The second assignment will also get id=1
3. process 1 deletes assignemnt(id=1)
4. Whoops, wrong assignment gets deleted

There are several options to resolve this problem:

1. Never reuse IDs. This would be easy, we do not need to modify the client code, but:
   1. The IDs will get reused eventually when the counter overflows
   2. User may chose IDs and therefore force reuse (although this may not really be a problem)
2. Forget about IDs for delete deltas and always delete by value. But:
   1. There may be value matching and authorization issues (e.g. metadata)