Initial support for key expire times with millisecond resolution. RDB version is now 3, new opcoded added for high resolution times. Redis is still able to correctly load RDB version 2. Tests passing but still a work in progress. API to specify milliseconds expires still missing, but the precision of normal expires is now already improved and working.

FLUSHALL now prevents rdbSave() from resetting the dirty counter, so that the command will get replicated and put inside the AOF. This fixes issue #142

don't process EXPIRE with negative TTL or EXPIREAT with time in the past if we are a slave too (see http://groups.google.com/group/redis-db/browse_thread/thread/5a931fefb88b16d5). Also propagate it as DEL.

They will be expired (and a DEL will be logged) after the loading is done
instead.

Fixed semantics of CLUSTER SETSLOT, SELECT now only denied in cluster mode if selected DB is not 0 so that MIGRATE still works well.

implemented a different approach to IO scheduling, so object->storage is no longer used, instead there is a queue and hash table of IO tasks to process, and it is always possible to know what are the scheduled and acrtive IO operations against every single key.

touched key for WATCH refactored into a more general thing that can be used also for the cache system. Some more changes towards diskstore working.