removing stale replication targets

[Andrew Tridgell]

Hi Matthieu,

I've just reverted your patch 5bfd6251. It was causing regular
segmentation faults in the build farm. As I said in my revert message, I
wasn't able to reproduce the segfault, but it is pretty clear that this
patch was the cause, given the backtraces and the complexity of the list
handling. What I think was happening was a DSA that had an in-progress
operation was being freed by the changes in this patch.

I'm also curious why the change was needed. The
dreplsrv_refresh_partitions() code should already be coping with changes
made by the KCC, including removals. Can you explain what situation this
was trying to cope with? If the patch is needed then we'll need to work
out how to make it work without the segfaults :-)

Cheers, Tridge