Deadlock when starting the server when using cache with clean_timeout=0

I'm using a configuration file with a cache.clean_timeout equal to 0. When I start the server and I specify on the command line the database to use a deadlock occurs.

This is due to the changeset a32b84d96649 that fixes a issue with testing. It results into the database lock used by the pool that is acquired twice (once when calling Pool.init, once when the transaction is started and it calls Cache.sync which call Pool.refresh, Pool.stop and tries to acquire again the lock (but it cannot because it's another thread).

I will probably move the while into the test case so that we're waiting there instead.