The python hooks from WT-7644 should call flush_tier() more often. An easy way to do this would be to add a hook that calls flush_tier() before connection.close().

This is easy to accomplish in principle.  But when I tried to add this to the work in WT-7644, lots of things broke.  For example the base01 tests started aborting.  

This ticket is to understand these failures and get the hook to work.  If this is tickling a bug in tiered storage, we'll have to fix that first.